A
REFERENCE GRAMMAR
OF KOREAN
A Complete Guide
to the Grammar and History
of the Korean Language
SAMUEL E. MARTIN
Yale University
CHARLES E. TUTTLE COMPANY
Rutland, Vermont & Tokyo, Japan
Published by the Charles E. Tuttle Company, Inc.
of Rutland, Vermont and Tokyo, Japan,
with editorial offices at
Suido 1-chome, 2-6, Bunkyo-ku, Tokyo 112
© 1992 by Charles E. Tuttle Publishing Co., Inc.
All rights reserved
ISBN: 0-8048-1887-8
LCC Card No.: 92-62395
First edition, 1992
Printed in Japan
CONTENTS
Part I. Korean Structure
0.0. Introduction. .I
0.1. Background and acknowledgments. .I
0.2. The structure of the book. .2
0.3. Orientation. .3
0.4. Grammatical terms. .3
0.5. Citations. .4
0.6. Romanization. .5
0.7. Arbitrary conventions. ...-I-.5
1.0. Letters. .6
1.1. Hankul symbols. .6
1.2. Hankul spelling. .7
1.3. Yale Romanization. .8
1.4. Transliteration rules. .9
1.5. Reinforcement (-q). .12
1.6. Initial 1 and n . .15
1.7. Hankul spelling of u after labials. .18
1.8. Word division and internal punctuation. .18
1.9. External punctuation. .20
1.10. Alphabetization. .21
2.0. Sounds. .23
2.1. Phonemes and components. .23
2.2. Vowel descriptions. .24
2.3. The pseudo-vowel uy. .26
2.4. Consonant descriptions. .27
2.5. Syllable structure and consonant liaison. .29
2.6. Cluster restrictions. .30
2.7. Sequence variants. .32
2.7.1. Precision variants. .32
2.7.2. Vowel length variants. .32
2.7.3. Disappearing h. .35
2.7.4. Disappearing w. .36
2.7.5. Postvocalic u. .37
2.7.6. Intercalated semivowels. .37
2.7.7. Desyllabification of i, wu, and o. .37
v ' Contents
2.7.8. Reduction of wie. . 38
2.7.9. Vowel assimilation. . 38
2.8. Standardization variants. . 39
2.9. Intonations. . 41
2.10. The earlier phonology. .42
2.10.1. The earlier vowels. .42
2.10.2. The earlier initials. . 43
2.10.3. Palatalization and dispalatalization. . 45
2.10.4. Nasal epenthesis. . 48
2.10.5. The earlier finals. . 49
2.10.6. Intersyllabic strings; assimilations; conflation and compression. .51
2.11. Lenitions and elisions; sources of G. . 53
2.11.1. Velar lenition and elision. . 54
2.11.2. Labial lenition and elision. . 56
2.11.3. Apical lenition; elisions of 1 and n. . 57
2.11.4. Sibilant lenition and elision. . 59
2.12. The accent of earlier forms. .60
2.12.1. Accentual patterns. .62
2.12.2. Accentual variants. .64
2.12.3. Accent suppression before particles. . 68
2.12.4. The accentuation of verb forms. .69
2.12.4.1. Vowel-final monosyllabic stems that are rising. . 73
2.12.4.2. Vowel-final monosyllabic stems that are low. . 75
2.12.4.3. Vowel-final monosyllabic stems that are high. .76
2.12.4.4. Vowel-final monosyllabic stems that are high/low. . 79
2.12.4.5. Bound stems. .84
2.12.5. Accent and spelling in Middle Korean texts. .85
3.0. Words. . 86
3.1. Inflected and uninflected words. . 86
3.2. Parts of speech. . 88
3.3. Free and bound words. . 88
3.4. Ionized parts of speech. . 88
3.5. Shortened words. .89
Chart: Parts of Speech. .90
3.6. Vocabulary. . 94
3.7. Layers of vocabulary in earlier Korean. . 95
4.0. Shapes. .98
4.1. Shape types. . 99
4.2. Syllable excess. .100
List of morph-final strings. .101
Examples of extra-syllabic excess. .103
4.3. Treatment of yey. .109
4.4. Alternations of 1 and n. . 110
4.5. Shape types of Chinese vocabulary. . 112
4.6. Chinese characters. . 113
Table: Shapes of Chinese morphemes. . 114
4.7. Characters with multiple readings. .116
4.7.1. Multiple readings: list one. .116
4.7.2. Multiple readings: list two. .120
4.7.3. Multiple readings: list three. .122
Contents vii
4.7.4. Index to the lists of multiple readings. .123
4.8. Chinese morphemes with basic l™. .124
4.9. Tongkwuk readings. .126
5.0. Forms: nouns. .130
5.1. One-shape and two-shape elements .130
5.2. Nouns. .130
5.2.1. Quasi-free nouns. .131
5.2.2. Free nouns. .131
5.2.3. Proper nouns; names and titles. .132
5.2.4. Deictics. .133
5.2.5. Adverbs. .135
5.2.6. Bound adverbs (preverbs, verb prefixes). .140
5.2.7. Interjections. .142
5.2.8. Bound nouns. .144
5.2.9. Bound preparticle. .145
5.3. Adnouns and pseudo-adnouns. .146
5.3.1. Quasi-adnouns. .150
5.3.2. Numerals. See §5.5.1.
5.3.3. Bound adnouns (prefixes). .151
5.4. Postnouns. .156
5.4.1. Postnoun/postmodifier adjectival noun (= adjectival postnoun/postmodifier). .160
5.4.2. Postsubstantives. .160
5.4.2.1. Postsubstantive adjectival noun. .160
5.4.3. Postmodifiers. .160
5.4.3.1. Postmodifier verbal noun intransitive (= inseparable
adjectival postsubstantive). .161
5.4.3.2. Postmodifier adjectival nouns inseparable
(= inseparable adjectival postmodifiers). .161
5.4.3.3. Postmodifier adjectival nouns separable
(= separable adjectival postmodifiers). .161
5.4.3.4. Pre-inseparable postmodifier. .161
5.4.4. Counters. See §5.5.3.
5.4.5. Bound postnouns (suffixes). .162
5.4.5.1. Core suffixes. .-•••162
5.4.5.2. Chinese suffixes. .165
5.5. Numbers. .171
5.5.1. Number constructions. .171
5.5.2. Numerals. .174
5.5.3. Counters. .179
5.5.4. Irregular counting. .185
5.5.5. Fractions. .188
5.6. Verbal nouns. .188
5.6.1. Defective verbal nouns. .189
5.6.2. Transitive verbal nouns. .190
5.6.3. Intransitive verbal nouns. .190
5.6.4. Adjectival nouns. .190
5.6.5. Bound adjectival nouns. .191
5.6.6. Conversion constraints on verbal nouns. .191
6.0. Forms: particles. .192
6.1. Characteristics of particles. . 192
v 'i* Contents
6.2. Quasi-particles. ..
6.3. Extended particle phrases (phrasal postpositions). . 194
6.4. Particles proper. ..
6.5. Particle sequences. . 197
6.5.1. List of particle sequences arranged by prior member. . 199
6.5.2. List of particle sequences arranged by latter member. .206
6.6. Sequences of ending + particle. .213
6.7. Some consequences of particle distribution. .214
Table of verb endings + particles and quasi-particles. .215
7.0. Forms: verbs. .216
7.1. Kinds of verbs. .216
7.2. Bound verbs. .219
7.2.1. Defective infinitives. .219
7.2.2. Bound adjectives. .220
7.2.3. Bound postverbs. .220
7.3. Defective verbs. .220
7.4. Causative and passive verbs. .221
7.5. Auxiliary verbs. .226
7.6. Postnominal verbs. .228
7.7. Recursiveness of auxiliary conversions. .229
Chart of double infinitive-auxiliary conversions. .230
8.0. Stems. .230
8.1. Conjugations. .230
8.2. Consonant stems. .231
8.2.1. Stems ending in sonants. .231
8.2.2. Stems ending in h. .232
8.2.3. Stems ending in w: -w- (= -p/w-). .233
8.2.4. Stems ending in 1: -T/L-. .234
8.2.5. S-dropping stems: -(s)-. .236
8.3. Vowel stems. .237
8.3.1. L-doubling vowel stems: -LL-. .238
8.3.2. L-extending vowel stems: -L-. .240
8.3.3. L-inserting vowel stems. .242
8.3.4. Ambivalent stems: -(H)-. .242
8.3.5. Irregular stems: ha- and derivatives. .243
8.3.6. Irregular stems: k-inserting and n-inserting. .243
9.0. Endings. .244
9.1. Sequence positions. .244
Table of endings. .246
9.2. Assertive and attentive endings. .248
9.3. Modifier endings. .249
9.4. The infinitive. .251
9.5. Substantives and derived substantives. .254
9.6. Derived adverb-noun forms. .255
9.7. Complex moods. .257
9.7.1. Complex moods built on the prospective modifier. .257
9.7.2. Adjunctives. .258
9.7.3. Complex moods built on the effective formative - 'ke-. .258
9.7.4. Gerund-related pseudo-moods. .259
9.8. Transferentives. .260
Contents ix
9.9. The structure of earlier verb endings. .260
9.9.1. Middle Korean finite forms: the basic scheme. .261
9.9.2. The effective. .262
9.9.3. Emotives. .263
9.9.4. Sentence types. .263
9.9.5. Aspect marking of sentence types. .265
9.9.6. Nonfinite endings. .265
9.9.7. Nominalizers. .267
9.9.8. Exaltation; the politeness marker (- ngi -). .268
9.9.9. The deferential zop -). .268
9.9.10. The modulator (-'w%-). .269
9.9.11. The copula. .273
10.0. Constructions. .274
10.1. Problems of word division. .274
10.2. Constructions and pseudo-constructions. .275
10.3. Compounds and quasi-compounds. .278
10.4. Phrases. .280
10.5. Sentences. .281
10.6. Sentences with multiple subjects and objects. .284
10.7. Other views of Korean syntax. .286
10.8. Syntactic constraints. .287
10.8.1. Subject-object expansion constraints. .287
10.8.2. Negative constraints. .289
10.8.3. Active adjectives; resultative verbs. .289
10.8.4. Constraints on modifiers. .290
10.8.5. Auxiliary constraints. .290
10.8.6. Emotive adjectives. .291
10.8.7. Separability constraints; auxiliary preemphasis. . 291
10.8.8. Animate-inanimate constraints. .291
10.8.9. Indirect-object intensification. .292
10.8.10. Locative constraints. .294
10.8.11. Copula and particle constraints. .295
10.8.12. Miscellaneous constraints. .295
11.0. Conversions. .296
11.1. Nuclear sentences and converted sentences. .296
11.2. Status conversions. .298
11.3. Style conversions. .299
11.3.1. Casual sentences. .301
11.3.2. Exclamatory sentences. .302
11.3.3. Circumstantial sentences. .303
11.3.4. Uncertainty sentences. .303
11.3.5. Afterthought sentences. .303
11.4. Tense-aspect conversions. . 304
11.5. Mood conversions. .305
11.5.1. The plain style. .305
Mood shift table. .306
11.5.2. The familiar style. .307
11.5.3. The intimate style. .308
11.5.4. The casual intimate style. .308
11.5.5. The casual polite style. . 309
X
Contents
11.5.6. The semiformal (authoritative) style. . 309
11.5.7. The polite style. .310
11.5.8. The formal style. . 311
11.6. Voice conversions. . 312
11.7. Negation conversions. . 315
11.7.1. Negatives and strong negatives. . 315
11.7.2. Negative preemphasis. . 316
11.7.3. Suppletive negatives. . 318
11.7.4. Negative commands and propositions. .320
11.7.5. Negatives with verbal nouns. . 321
11.7.6. Double negatives. .321
11.7.7. Other negative expressions. .322
11.7.8. Negative sentences with positive force. .322
11.8. Nominalizations. .323
11.9. Adnominalizations; epithemes. .324
11.10. Adverbializations. .329
11.11. Quotations; oblique questions; putative structures. . 331
11.12. Reflexive requests; favors. . 333
11.13. Sentence connectors. . 333
11.14. Apposition. . 335
11.15. Order and recurrence of conversions. . 335
11.16. Sentence generation. .336
12.0. Mimetics. .340
12.1. Phonetic symbolism. .340
12.2. Phonomimes and phenomimes. .340
12.3. Intensives and paraintensives. . 343
12.4. Word isotopes. . 343
12.5. Mimetic constructions. . 344
12.6. Shapes of mimetic adverbs. . 346
12.7. Iteration. . 347
Appendix 1. Lists of stem shapes. . 348
Appendix 2. Korean surnames. .366
Appendix 3. Korean provinces. .370
Appendix 4. Japanese placenames. . 371
Appendix 5. Radical names. .372
Appendix 6 . List of Korean grammar terms. .380
Appendix 7. English index to the list of Korean grammar terms. .389
Appendix 8 . Chronological list of texts. . 397
Appendix 9. Alphabetical list of texts. .401
Bibliography. .407
Table of abbreviations. .414
Part II. Grammatical Lexicon .415
Index .957
PART 1
KOREAN STRUCTURE
V,
A Reference Grammar of Korean
PARTI 1
0.0. Introduction.
A Reference Grammar of Korean is a description of the language spoken in both north and south
Korea in the second half of the 20th century. This material is given historical perspective by a
description of the structure of the language of the Hankul texts of the second half of the fifteenth
century and somewhat later, here called Middle Korean (MK); occasional reference is made to still
earlier forms of the language, for which we have only very limited materials in the form of Chinese
characters used for their sound value (i.e. as phonograms) or inferences that are made from systematic
irregularities in the grammar of the earliest Hankul texts. Attention is paid also to dialect variation
reported for the modern language and earlier speech. The core of the modern material reflects the
pronunciation and usage of speakers who were born in Seoul before 1950, but that has been updated by
observations of the speech habits of younger speakers, both in Seoul and elsewhere. In addition to
direct elicitation, tape recordings, and written materials, VCR tapes of quite recent Seoul TV programs
have provided authentic data of the contemporary usages referred to in various sections of the book.
The sentences are presented in a Romanized form which can be readily converted to a Hankul
representation, but they are intended to write spoken language, and do not always coincide with the
prescribed spellings. The reader should be aware that Romanized forms such as pat.e and iss.ey yo are
not mistakes, but represent the relaxed pronunciation of the Seoul speakers who have provided or
checked them, though the speakers would indeed themselves write the words with the usually seen
spellings “pat.a” and “iss.e yo”, following the pronunciations heard in other areas and increasingly
among younger speakers in Seoul.
Although a good deal is said about the history of sounds and forms, this work does not address
questions of prehistory or genetic relationships. When references are made to Japanese, Chinese, or
English data the intention is usually to show how the Korean counterparts are similar or different, or to
add perspective on the meaning.
0.1. Background and acknowledgments.
This book has been put together over a period of more than forty years. The first version was
written in 1960 under the Program in Uralic and Altaic Languages of the American Council of Learned
Societies with support from the U.S. Office of Education. That version was later made available for a
time through Bell-Howell. Meanwhile, after a trip to Korea in 1960, I set to work making many
revisions and additions, resulting in the 1963 version, which I had hoped to turn into publishable form
in fairly short order. The project was set aside, however, because of other priorities: publication of the
Korean-English Dictionary and of the textbook Beginning Korean. Then I became deeply involved in
writing A Reference Grammar of Japanese, which occupied my time and thoughts for eight years. In
doing the research for that book I came to realize the inadequacies of the work I had done on Korean,
especially in view of new ideas on Korean syntax which had appeared. Despite that, I decided to
circulate photocopies of the 1963 version to a few colleagues for their comments and to use in Yale
University seminars in the structure of the Korean language. As a result of similar seminars in the
history of Korean, I came to feel the necessity of including materials from earlier centuries which
provide perspective on the modern language, and that is what led to the dual nature of the book you
see, for it attempts to set the synchronic description into its historical background, which often sheds
revealing light on vexing problems.
While many of the example sentences are taken from published texts, most of the modern
examples were elicited from Korean colleagues and informants, who were generous with their time and
knowledge. I am particularly grateful to Sung-Un Chang (Cang Sengen) and Young-Sook Chang Lee
([‘Yi] Cang Yengswuk) who provided perception and insight, as well as many of the best examples; the
late scholar Yang Ha Lee ('Yi Yangha) was helpful during his collaboration with Sung-Un Chang and
me while we were compiling our Korean-English Dictionary. In connection with another project I was
able to elicit examples from Sek Yen [Kim] Cho ([Co] Kim Sek.yen) which were useful for this book. I
have built upon earlier linguistic work published by Fred Lukoff and Elinor Clarke Horne, and I have
freely incorporated material from their books. I have culled good examples from teaching materials
2 PARTI
A Reference Grammar of Korean
prepared by Edward W. Wagner and the excellent textbooks of the Myongdo Language Institute
prepared by A.V. Vandesande and colleagues. I have also used to advantage materials found in works
by linguists in Korea, both the south and the north. In 1961 I enjoyed some excellent discussions with,
among others, Woong Huh (He Wung), Hie-seung Lee ('Yi Huysung), Nam Tuk Lee ('Yi Namtek),
Sung Ny(e)ong Lee ('Yi Swungnyeng), Chang-Hai Park (Pak Changhay), and Bong Nam Park (Pak
Pongnam), whose textbook for the Foreign Service Institute was helpful. I am grateful to Fred Lukoff
for introducing me to MaengSung Lee ('Yi Mayngseng), whose assistance in checking delicate points
of syntax was of great value. I learned much from my students, not only at Yale but at the University
of Washington and at the University of Hawaii, many of whom have become eminent scholars and
teachers. In particular, I have been much helped by the stimulating ideas through the years of S.
Robert Ramsey and, more recently, J. Ross King; both have been excellent critics who shaped my
thinking in many ways. Seungja Choi (Choy Sungca) and Sun-Hee Kim (Kim Senhuy) have provided
excellent observations and examples. The final manuscript was read by Choi, King, and Ramsey, who
suggested many corrections and improvements, most of which I was able to include. I began this final
revision of the work in 1989, while on a sabbatical term at the Center for Korean Studies of the
University of Hawaii, where both Dong-Jae Lee ('Yi Tongcay) and Ho-min Sohn (Son Homin) gave
generous consultation and assistance. I am grateful to them and also to other members of the Center
and of the Department of East Asian Languages and the Department of Linguistics for the various
facilities they provided. Byron Bender, Robert Hsu, G.B. Mathias, Albert Schiitz, and J.M. Unger
helped make that half year both productive and pleasant. A great many other scholars of the past and
the present, including the late Hyon-Pai Choi (Choy Hyenpay), contributed to the ideas in this book
through their published writings and private discussions, and their names will be found throughout the
book and in the bibliography. I would be remiss not to mention my indebtedness for information on the
modern language to Suk-Jin Chang (Cang Sekcin), Choy Hak.kun, Min-Soo Kim (Kim Minswu),
Young-Key Kim-Renaud (Kim Yengki), Kim Thaykyun, Ko Yengkun, Ki-Shim Nam (Nam Kisim),
Seok-Choong Song [Song Sekcwung], In-Seok Yang ('Yang Insek), Joe Jung-No Ree ('Yi Cenglo), Ki-
dong Lee ('Yi Kitong), and many others mentioned in the following pages. For information on the
older language I appreciate the fine work of Huh Woong (He Wung), Wan-jin Kim (Kim Wancin), Kim
Minswu, Gwang U Nam (Nam Kwangwu), Kono Rokuro, Ki-Moon Lee ('Yi Kimun), and Sung
Ny(e)ong Lee ('Yi Swungnyeng), among others, and I hold ever deepening respect for the prodigious
achievements of the late Chang-Ton Yu ('Yu Changton).
In citing personal names here and in the bibliography I have tried to include the Romanized form
preferred by the person, when that is known to me, accompanied by a consistent version in the Yale
Romanization. When information on personal preference was not available to me, I cite only the Yale
version. To insure consistency, the Yale form is generally used in references within the book.
The preparation of the camera-ready copy of this book was much facilitated by the technical
expertise and wisdom of my colleague and mentor Rufus S. Hendon, who has been helpful at every
step of the way. His willingness to create and share software to answer my needs is deeply appreciated,
as is his patient guidance through difficult problems, where his advice has been unfailingly sound.
0.2. The structure of the book.
A Reference Grammar of Korean is divided into two parts. Part I is a systematic survey of the
structure, in which we examine problems of orthography and grammar, set up a system of parts of
speech, analyze the constituents of sentences, and explore systematic relationships between sentences.
Part II is a grammatical lexicon, an alphabetically arranged list of particles, endings, affixes,
auxiliary verbs, and other grammatically interesting elements, along with certain additional words
(including ordinary nouns and verbs) to which quick reference may help clarify the other words listed.
This part is not a substitute for a dictionary, since it does not contain most of the “content” words of
the lexicon. It was my intention to make the list so complete that the user would find sufficient
information about each element of a sentence, other than the meaning of nouns and verb stems, to
figure out the grammar of any sentence and be able to translate it accurately.
A Reference Grammar of Korean
PART I 3
The various appendix lists are intended to help the user find the meaning or shape of terms not
easily located elsewhere, as well as other information that is relatively inaccessible in other sources.
0.3. Orientation.
This book is not trying to prove a theory about the nature of language. 1 do not maintain that the
structure of a language is either discoverable or describable in one and only one “correct”, or even
uniquely “best”, way. The criteria for judging a description vary with the purpose for which it is
intended. For a reference grammar the most important criterion is balanced completeness. As much
useful information as possible must be given in a form that makes it readily accessible to the user. The
information that is most often, or most sorely, needed should be the easiest to get at. Lists are not to
be scorned; formulas are not to be worshipped. Economy of statement is a technical criterion relevant
to the accessibility of the information; elegance of statement is a psychological criterion relevant to the
impact of the information.
Just what information is useful and for whom? The foreigner who is decoding (making out the
meaning of) messages spoken or written in Korean is concerned, first of all, with the constituency
of sentences. Given a sentence, he wants to know what are its pieces and how do they fit together. He
needs to be able to take the sentence apart, to “parse” it. The foreigner who is encoding (making up)
messages in Korean is interested in the manipulation of sentences. Given known sentences, what
new but related sentences can he say that will be understood and accepted by Koreans? Up to the
sentence level our presentation is in terms of item and arrangement: the items are morphemes (words
or parts of words), and the arrangement is stated in terms of immediate constituents (IC’s) or, in a few
cases, unordered strings. Beyond the sentence level, the presentation is in terms of item and process:
the items are certain types of simplex sentences, and the processes are conversions that turn these
into more complex structures.
0.4. Grammatical terms.
You may find the terminology unfamiliar and irksome. If so, think of the categories in terms of
concrete representations. Should you not feel comfortable referring to a common form of the verb as
the “gerund”, take that just to mean “the -ko form” of that verb - or, if you prefer, “the hako form”.
The grammatical categories of Korean are numerous, diverse, and complexly represented. They cannot
easily be put into a frame of reference based on the descriptions of other languages. But in practice it is
convenient to choose terms that are somewhat familiar, supplemented when necessary by new terms
made up by analogy, with the clear understanding that NO direct correspondence is intended with
the categories of other languages that are given similar names.
The set of names found in this book has grown out of terms used in earlier books; many of them
stem from Elinor Clark Horne and our teacher Bernard Bloch. One that has troubled many people is
“infinitive” for the -e ending. Regardless of the merits of the word itself, the name has become so
widespread in discussions of the grammar of Japanese (where it refers to the m form, sometimes called
“continuative stem”) that it has surely become the standard. For what is here called the “summative” I
earlier applied Horne’s term “nominative”, but that is better used in reference to the case-marking
function of the particle - i/ka, and so I have abandoned its use for any other purpose. The word
“substantive” is sometimes used as a general term for “noun”, but here it is narrowed in definition to
one of the endings (-um) that make the verb into a form that is used like a noun; other such forms are
the summative -ki and the “derived noun” -i. For the -key ending I have changed the earlier term
“adverbial” to “adverbative” so that “adverbial” can be used to discuss syntactic phenomena only. I
adopt Wagner’s term “purposive” to refer to the -ule ending, but what he calls the “expository” is here
called the “sequential” (the -uni ending), and what he called the “effective” is called the “projective”
(the -tolok ending), so that the term “effective” can designate an aspect marker of Middle Korean
(-7A- or -'%-). In speaking of sentence styles I continue to use “authoritative” to refer to verb forms
ending in -o/-so (etc.), as a synonym for “semiformal”, Wagner’s term, which characterizes the style
in a broader way.
4 PARTI
A Reference Grammar of Korean
0.5. Citations.
Examples taken from modern written materials are not always attributed to the source, since they
have frequently been edited as the result of elicitation. Examples from earlier texts are cited by date,
text, and page. Two lists of these texts are appended: one arranged chronologically, the other
alphabetically. Some of the dates are questionable; I have done my best to make practical decisions on
the basis of the bibliographical materials available to me, and to add question marks when they seem
appropriate. The intention is to give a specific date whenever possible.
The translations of the text sentences are mostly my own, sometimes made in consultation with
others. I believe they are adequate to convey the meanings of the grammatical structures, but I have
made only a limited effort to check the translation of philosophical concepts involved in Buddhism or
references to ancient China. Chinese names, whether modern or ancient, are given in the now standard
Pinyin Romanization. References to Middle Chinese forms of the 7th century follow the notations used
for similar purposes in Martin 1987 and represent rough approximations to the pronunciation of
northern China in the 7th century, along the lines of the phonemic analysis in Martin 1953.
Modern Korean forms are printed in boldface; forms of earlier Korean are printed in italics. The
handling of vowel distinctions is slightly different in the Romanization as used to represent the pre¬
modern spellings and that used for the modern language. In citing pre-1933 spellings which retain
obsolete features (such as the low back vowel o that mostly became a- and -u- in Seoul) the notation
writes wu and wo for all cases of the back rounded vowels, u and o for all cases of the back
unrounded. This “expansive” notation is shown in the italic font used for the Middle Korean forms.
Forms from Ceycwu island, which retains the low back vowel (pronouncing it as rounded but distinct
from wo), are cited in the same way. Unless otherwise specified, an italicized word is to be taken as
Middle Korean. When a Chinese character in a Middle Korean text is accompanied by a Hankul
syllable to show the prescribed readings, that syllable is printed in small italic capitals. When the
character fails to carry a notation of the reading (as in 1481 Twusi) we get the reading from the
prescriptions in 1447 Tongkwuk cengwun and put brackets around the word; if the reading is
unavailable in Tongkwuk, we infer it from other sources and put an asterisk before the string of small
capitals used to represent the syllable.
The quotation marks around forms cited from pre-Hankul sources are intended to remind us that
the phonograms are interpreted faute de mieux in terms of their mid-15th century Hankul values; the
semantograms (characters used for their meanings) are given as small-capital English words. The Yale
Romanization is used for Japanese words, which are underlined. Sanskrit and Chinese words are not
typographically differentiated from English, but the diacritic marks usually make them obvious.
Examples were chosen purely to illustrate structural patterns, and I have made no effort to alter
ideological, religious, or sexist content that may seem quaint or even offensive. Notice that the term
“vulgar” as used here does not mean obscene; rather, it refers to quite colloquial forms, including
slang, that are generally avoided in writing and in less relaxed speech.
References to North Korean data are mostly taken from Cosen mal sacen (NKd) and Cosen-e
munqpep (CM). As Kim Minswu 1985 points out, there were three distinct periods of prescriptive
standardization in North Korea, referred to by the names of the authoritative publications: 1945-54 (the
Thongil-an period), 1954-66 (the Chelqcaq-pep period) and 1966- (the Kyupem-cip period). In 1954,
after using the unified spelling system that dates from the 1933 Thongil-an, North Korea published
Cosen chelqcaq-pep, which introduced a number of changes, such as using the apostrophe for the sai
phyo that is here Romanized as —q (§1.5). Some of the changes were abandoned, in whole or in part,
with publication of the 1966 Cosen mal kyupem cip, which made efforts to create a normalized
“munhwa-e” (cultural language) that incorporated a few dialect or outdated elements (Kim Minswu
1985:129), including iya yo for the polite copula that is treated as iey yo in modern Seoul. And the
1966 rules prescribed the artificial pronunciation of /I/ (as a flap r) rather than Ini for the initial of
words here Romanized as 1 n~ which come from Chinese 1-, including *i(—) and *y~, where the initial
is elided in the south and was at one time pronounced with n- in the north. (See §1.6.)
A Reference Grammar of Korean
PARTI 5
0.6. Romanization.
There are a number of systems for writing Korean words in Romanized form, depending on the
purposes for doing so. The popular McCune-Reischauer Romanization, which received official sanction
from the Ministry of Education in 1988, tries to approximate the way a Korean word sounds to the
American ear, disregarding its internal structure and history. It is generally preferred in citing Korean
names and in casually mentioning Korean words in English context. The Yale Romanization, like the
various systems of Hankul spelling, takes account of more than just the sound. The two given names
Pok.nam and Pongnam are both written “Pongnam” in the McCune-Reischauer system, but they are
kept distinct both in Hankul and in the Yale Romanization. The two surnames Yang and ‘Yang are
spelled with the Hankul equivalents of “Yang” and “Lyang” in North Korea, but both are written
“Yang” in South Korea and in the McCune-Reischauer Romanization. Four names can be shown as
different by the Yale Romanization (Yang Pok.nam, Yang Pongnam, ‘Yang Pok.nam, ‘Yang
Pongnam - and four more if there are people named Pok.lam and Pong.lam) but will all be treated
alike by the McCune-Reischauer spelling “Yang Pongnam” (or “Yang Bongnam” if the space is
ignored). Numerous problems of detail have to be handled arbitrarily in each system of spelling,
whether it be Hankul or Romanization. The forms cited for the modern language in this book are not
merely mechanical transliterations of Hankul spellings in one or another orthographic standard, but
offer additional information about the background of the forms, including phonological details often
ignored in Hankul. Koreans may be particularly irritated by the generous sprinkling of “~q” to mark
certain reinforcements that go unmarked in the usual Hankul spellings. In particular, notations
reflecting what is sometimes called “n-epenthesis” (see §1.5), such as mosq ip.e and anq ip.e or mosq
yel.e and anq yel.e, may strike the eye as unnecessary nuisances, since the phenomenon they represent
is not immediately apparent to the naive ear. The current trend in Hankul orthography (especially for
words with Chinese components such as munqpep ‘grammar’) is to ignore most of the cases of
reinforcement, including many of those which come from the Middle Korean adnominal particle s.
Perhaps it would be more congenial if we made the notation smaller or subscript (mos,, ip.e, an,, ip.e)
or used a flimsier symbol, such as the apostrophe, here unavailable because it is needed for other
purposes. But we will retain the full notation and invite readers to ignore it when that seems
appropriate, as this book often does in alphabetizing lists.
0.7. Arbitrary conventions.
The notational devices and the decisions on punctuation, capitalization, and word division will
strike you at times as needlessly fussy or cranky, and of less interest to the student than they are to the
technical linguist. Feel free to simplify, modify, and adapt the notations to your own needs. I have
attempted to present the material in a manner intended to be maximally useful, one that can be readily
converted to that of other systems which retain less information. Decisions on the use of hyphens,
apostrophes, and spaces may seem arbitrary in particular cases, as when we write “na-ka-, na-o-, na-
su-” and “ka-po-” (but “wa po-”); they are intended to make it easier to identify the phrases. The
apostrophe shows where a sound or a string of sounds is omitted in an abbreviation, as in kanta
’yss.ta < kanta (ko) hayss.ta ‘said (that) he was going’.
Sentences of Romanized Korean, like those of English, begin with capitalized letters, as do names.
But both the citation and the translation of the Middle Korean and Ceycwu examples begin
uncapitalized. In the italic notations, W and G represent the voiced fricatives [ft\ and [Y]. (We have
chosen not to use lowercase v and g, in part because of fear that they would be susceptible to
misinterpretation, though these letters would be consistent with our use of z for the voiced sibilant.)
The capital letters “C” and “V” are used in formulas to mean “(any) consonant” and “(any) vowel”; in
other contexts “v” or “V” represents “verb (stem/form)”. For both varieties of the Yale Romanization
we use the conventional digraph “ng” to represent the velar nasal, rather than the single symbol
provided by the phonetic notation [q], Specifically phonetic notations are usually put between brackets.
Specifically phonemic notations are in boldface or italic between slashes, as in “/ng/ and Ingl are
pronounced as the velar nasal [q]”. Brackets are also used to demarcate ellipted words, or parts of
6 PARTI
A Reference Grammar of Korean
words, and elisions of a phoneme or a string of phonemes. When the slash is between two words or
letters, it has the usual meaning of “or (optionally)”. Material within parentheses may be either
enlargements or optional replacements; the context should make the intention clear.
When a Korean citation is within an English sentence, the gloss (translation) is set off with single
quotation marks, as in “The word koki ‘meat’, for example, ... ”, but the marks are omitted for the
glosses of examples not embedded in such context, since the demarcations are obvious.
The macron is used to mark long vowels in modern words like ton ‘money’, nwun ‘snow’, umsik
‘food’, sTl ‘thread’, sem ‘island’, and pam ‘chestnut’, though the distinction of long and short vowels
has been largely lost by younger Seoul speakers. The dots in the Middle Korean words represent the
distinctive pitch of the following syllable (one dot for high, two dots for low rise), and unmarked
syllables are treated as low, provided the text is one that normally marked the accent. The position of
the dots is kept just where it is in the original text, so that swon ‘guest’ + i [nominative particle] is
written swo'n i (1445 'Yong 28) and pronounced jvw™. The spaces and hyphens correspond to nothing
overt in the spellings or pronunciations; they are there just to help your eye identify words and parts of
words. MK nouns that appear in an environment where a basic final •••h after a vowel or a resonant is
suppressed (and not shown by the Hankul) are Romanized as -V[h], -l[h], -n[hj. But for the other
syllable-excess nouns (§4.2) no indication is given of the basic shape before the reduction: the shapes
are written according to the spellings in the texts, which most often indicate the syllables heard rather
than the underlying forms.
1.0. Letters.
Korean words can be spelled out with foreign letters of various kinds, including the familiar
Roman letters used in this book, or with letters created in Korea by King Seycong and promulgated in
1445. Words of Chinese origin are often written with Chinese characters either alone, without letters to
represent the pronunciation, or added in parentheses to identify a word difficult to understand from the
sound alone. In modern south Korea it is a matter of controversy whether the continued use of Chinese
characters should be encouraged or discouraged.
1.1. Hankul symbols.
Koreans usually write in an alphabet known as Hankul. As a result, the word Hankul is also used
to mean ‘Korean language, especially as written’ and ‘Korean letters = literature’. In the system of
writing there are symbols to represent each of the phonemes of Korean. The term “Hankul” was first
used by Cwu Sikyeng in 1910; earlier the symbols were called enmun. It is possible to use the system
as we use Roman letters, writing horizontally across the line letter by letter: that is called kalo-ssuki
‘horizontal writing’. But usually the symbols are joined into written syllables, which often but not
always correspond to spoken syllables. The written syllables are made up of an “initial” consonant
(including zero) + a vowel nucleus or “medial”, consisting of a vowel or a semivowel (y or w) + a
vowel, written as a unit. The vowel is sometimes followed by a “final” consonant or two-consonant
cluster that is called by the Korean grammarians pat.chim ‘pedestal’. The syllables are written as
blocks (called “logotypes” by 1893 Scott) designed to resemble the shape and spacing of Chinese
characters, which are still often used to write, or to clarify, Chinese words in the midst of native
Korean words. The initial is written at the top, the nucleus either on the right (those containing a or e
or i) or below (those containing u or wu or o). A final pat.chim is placed below everything - and
slightly to the right, if it is but a single consonant. The shapes of the symbols are altered a bit when
they appear in different positions. For example, before a the k swoops back to the left 7|- ka, but
astride the o it has a straighter fall HI ko, and the final k is longer, flatter, and straighter: mak.
Since there are a limited number of Hankul symbols, representing the basic phonemes of Korean,
we can substitute our Roman letters for the Hankul letters with no loss of information. If we are
consistent, the Roman transcription can be automatically converted into the Hankul version and vice
versa. There are several ways in which Korean can be Romanized, and each scheme involves certain
difficulties and special rules. The system used in this book takes time to get used to, because it is
A Reference Grammar of Korean
PART I 7
designed to be typographically simple by avoiding “odd” letters, such as 6 (= e) and u (= u). But
once you are familiar with the system, it is very easy to use; you will find it more flexible than other
systems of putting Korean into Roman letters. With this system anything written in Hankul can be
typed out on an ordinary English typewriter or computer keyboard with no special tricks.
How many different syllables does Korean have? For the modern language, computer codes allow
about 2,500 Hankul syllables to be differentiated. Using all conceivable sequences, including many that
do not occur in any word, ‘Yu Huy (1824 Enmun-ci 18) came up with the staggering number of 10,250
possible syllables. And Kim Hyenglyong (1985:31) found a total of 11,172 different orthographic (=
morphophonemic) syllables. His study (31-2) found the total number of different two-syllable strings
(morphophonemic dissyllables) to be 31,759,684, of these types (V includes yV):
Type
Example
Strings
1 .
V-V
oi
441
2.
V-VC
a.yang
11,907
3.
V-CV
oli
7,938
4.
V-CVC
yuceng
214,326
5.
CV-V
kyo.yey
7,938
6.
CV-CV
cwuchey
142,884
7.
CV-VC
kyoyang
214,326
8.
cv-cvc
pltan
3,857,868
9.
vc-v
Yenge
11,907
10.
vc-cv
ak.ki
214,326
11.
vc-vc
ek.yang
321,489
12.
vc-cvc
wuntong
5,186,802
13.
cvc-v
kak.o
214,326
14.
cvc-cv
sokto
3,857,868
15.
cvc-vc
cengqyel
5,186,802
16.
cvc-cvc
hyek.myeng
12,308,536
1.2. Hankul spelling.
Koreans, like speakers of English, have spelling problems. Although the Hankul system of writing
is very simple, it is not easy to devise a consistent system for spelling out the words of the language.
That is because the structure of Korean is somewhat complicated, in that words and parts of words
often change the way they sound depending on the words around them. The Korean writer has a
choice: he can use the Hankul symbols (or their Roman equivalents) to write phrases EXACTLY AS
they sound, or he can write individual words and parts of words always the same way regardless
of changes in sound. The first method, known as a PHONEMIC orthography, has the advantage that
even a foreigner beginning his study of Korean can read sentences without learning a lot of special
rules; and he can write down everything he hears - provided he hears accurately - without worrying
about what words the phrases contain. But the disadvantage to the reader who already knows the
language is obvious: the same word appears sometimes in one shape, sometimes in another. For that
reason, native speakers of Korean naturally prefer some sort of morphophonemic orthography, as
linguists call the second kind of spelling. Morphophonemic spelling tells the reader a lot about the
grammar of the phrase he is reading, since it tries to spell each word (or part of a word) always in the
same basic shape, with the expectation that the reader will be able to apply a set of rules that will
automatically produce the particular phonemic shape needed to pronounce the phrase. We do something
similar in English when we write our plural ending as “s” both after “cat”, where it is pronounced as
an s, and after “dog”, where it is pronounced as a z.
The difficulties that arise in using a morphophonemic orthography are of three kinds:
(1) How far should we go in analyzing words into parts? How can we be sure we have the “same”
word-part (= morpheme or string of morphemes) in different words? To what extent should we allow
our knowledge, or someone’s knowledge, of the history of the words to influence the decision?
8 PARTI
A Reference Grammar of Korean
(2) When there are several spoken variants for a word, should we try to settle on one as the
“standard” form and ignore the others when we write? Perhaps we can let two or more forms coexist
as model and shortening, as with the English forms “do not” and “don’t”; or as literary and
colloquial, like English “unto” and “to”; or as dialect and standard, like English “dreamt”
and “dreamed”, “dove” and “dived”.
(3) What specific spelling devices should we use to handle certain tricky problems, such as
reinforcement (the sai sios or -q phenomena, §1.5) or the complications of initial 1 and n which lead to
the use of the superscript letters 1 and n in the Yale Romanization (§1.6)?
Koreans have contended with these problems for many years, and there have been several attempts
to prescribe consistent and comprehensive spelling systems. Two spelling systems have come to enjoy
wide use in the years since the end of World War II. One is the official system of South Korea (the
Republic of Korea), sometimes called the Thongil-an or Unified System; the other is that of North
Korea (the Democratic People’s Republic of Korea or DPRK). The two systems are almost identical in
the way they treat problems of the first and second types; they differ in their approach to problems of
the third type. Since each system has some advantages, and both are widely used and sanctioned, the
spellings of this book are designed to convert into either system automatically. Both systems have
undergone several minor revisions and the usages reflected in this book may be modified in future
revisions of the systems.
1.3. Yale Romanization.
In the system of Yale Romanization used in this book, the dot or period is used within a word for
several purposes, some more important than others:
(1) The dot is used to indicate the “zero” (vowel) beginning of a syllable or other ambiguous
situations when the preceding Hankul syllable ends in a pat.chim (final) consonant: mek.e, mek.ko,
mek.hinta. Of course, when the boundary is shown by a space or a hyphen, the dot is unnecessary:
Puk Han ‘North Korea’, kak-kak “each separately”.
(2) The dot is also used to distinguish e.yV from eyV, ay.V from ayV, u.yV from uyV, and o.yV
from oyV. We use the digraphs ey, ay, uy, and oy as units except when a dot intervenes. The dot is
omitted, however, when the y follows yo because there is no string *yoy, so there is no possibility of
misinterpreting the syllable division in a word like kyoyuk ‘education’. In a word like mu.yek ‘trade’
the dot is just a reminder, because in modern Korean uy does not occur after a labial. In a word like
kup.wu ‘classmate’, the dot is not strictly necessary, since the syllable /pwu/ is simplified to the
spelling pu (modern Korean lacks the unrounded syllable */pu/), but I retain the dot for clarity.
(3) A third use of the dot is to remind the reader of the automatic morphophonemic rules (sound
changes) between two consonants: — th.t — is pronounced — tt—, -ch.s- is pronounced -ss-, etc.;
-p.m- is pronounced -mm-, -n.l— is pronounced —11—; in verb forms -n.t— is pronounced as if
—ntt—, — n.k— as if -nkk-, -n.c- as if -ncc-, -n.s- as if -nss™. If we were not trying to follow
the Hankul spelling systems, which ignore all three kinds of sound changes, we would spell the verb
forms -mqt-, -nqk™, -nqc™, -nqs-, etc.
In the third use, the dot can be omitted with no loss of essential information about Hankul spelling
or even (given knowledge in the last case that the word is a verb form) the ultimate pronunciation, and
that is what we do in using the Romanization in English contexts, such as the Bibliography. In the first
two uses the dot is essential to recover all the word-structure information contained in the Hankul
spelling; in the second use it is required as a result of choosing digraphs to write some of the vowels.
In later sections the sound system of Korean is discussed in terms of the Romanization, and the
several digraphs are treated as single units. The consonants are represented by p t c k s, pp tt cc kk
ss, ph th ch kh h, m n ng 1; the vowel nuclei are represented by i ey ay u e a wu (abbreviated to u
after a labial) o oy, yey yay ye ya yu (an abbreviation of ywu) yo, wi wey way we wa. In addition the
letter q is used as a special morphophonemic symbol to show reinforcement (see §1.5), and superscript
1 ™ and n — show differences between the spelling systems of the two Koreas (see §1.6). In this book
A Reference Grammar of Korean
PART I 9
the length of a long vowel, ignored in the usual Hankul spelling, is shown by a macron (a line above
the vowel); see §2.7.2.
1.4. Transliteration rules.
Here are the complete rules for transcribing Hankul into the Yale Romanization, with the
exception of the problems of 1 ™, and -q, which are covered in the following sections. The
Romanization uses 16 of the usual 26 Roman letters, plus the digraph ng, final -q, and initial
superscript 1 - and ”•••. You will occasionally find other superscripts used to show divergences between
the spellings of the two Koreas: s in iss. s o and iss. s up.nita; h in an h ay (see §2.8); y in ph y ey and
m y ey (see §4.3). In the table of rules the word “space” means “space, hyphen, or other punctuation”,
C means consonant, 0 means initial zero (the syllable starts with a vowel, y, or w).
CONSONANTS
Hankul
Initial
Example
Final
Examples
~)
k
kikwu
k(.) + m n 1
k. + k h 0
k
sik.mo (or sikmo)
kyek.nyen (or kyeknyen)
tok.lip (or toklip)
sik.kwu, kak.ha, mek.e
yak, siktang, *yukpun, mekca,
'yuksip, patak, ...
kh
khal
kh + space
kh(.) + C
kh. + 0
puekh
puekh.teyki (or puekhteyki)
puekh.an (but better written
as two words)
77
kk
kkay
kk + space
kk + 0 k (kh)
kk(.)
pakk
kkakk.un, kkakk.ko
kkakk.ta (or kkakkta),
kkakk.nun (or kkakknun)
U
P
papo
p(.) + mnl
p. + p h 0
P
ip.mun (or ipmun)
ip.nap (or ipnap)
sip.lyuk (or siplyuk)
kup.po, ip.hak, ip.e
ip, ipta, ipko, ipsang, ipcang,
H
ph
pha
ph + space
ph(.) + C
ph. + 0
iph
aph.cang (or aphcang)
aph.aph-i
WH
PP
ppye
—
—
C
t
titinta
t + space
t. + t h 0
t(.)
kot
tat.ta
tat.hinta
tat.un
tat.chinta (or tatchinta),
tat.ko (or tatko),
tat.nun (or tatnun), ...
E
th
thal
th + space
th. + 0
th(.) + C
path
puth.e
puth.ta (or puthta),
puth.ci (or puthci),
puth.nun (or puthnun), ...
10 PART I
Hankul
Initial
Example
Final
n
U
ttal
—
A
c
cal
c + space
c. 4- c h 0
c(.)
X
ch
cha
ch + space
ch(.) + 0
ch(.)
A
cc
ccanta
-
A
s
san
s + space
s. + s 0
s(.)
/A
ss
ssal
ss. + s 0
ss(.)
o
h
hay
h. + 0
h(.)
□
m
mal
m. + 0
m(.) + 1
m
(But Cf §8.1.1
L_
n
nal
n. + 0
n(.) + 1
n
(But Cf §8.1.1
e
1
latio
1. + 0
K.) + n
1
0
(ZERO)
al, yen,
wenca
ng(.) + 1
ng
S')
-
lk. + k h 0
lk(.)
A Reference Grammar of Korean
Examples
nac
mac.ci
mac.hinta
mac.un
mac.ta (or macta),
mac.chwunta (or macchwunta),
mac.nun (or macnun)
kkoch
coch.a
coch.ko (or cochko), coch.ci (or cochci),
coch.nun (or cochnun)
os
wus. s up.nita
wus.e
wus.ko (or wusko), wus.ci (or wusci),
wus.nun (or wusnun)
iss. s o
iss.e
iss.ta (or issta), iss.ko (or issko),
iss.nun (or issnun)
noh.a
noh.ko (or nohko), noh.ci (or nohci),
noh.nun (or nohnun)
sim. e
chim.lyak (or chimlyak)
nam, simpang, slmnun, chamko
slm.ko, slm.ta, slm.ci.)
sin. e
Sin.la (or Sinla), Cen.la (or Cenla)
an, sinnun, sinmun, cinpo,
"yento, "yenkam, mence, ...
sin.ko, sin.ta, sin.ci.)
sil.ep
il.nyen (or ilnyen)
kil, silkwa, silqswu, mullon,
ppalli, kolmu, kilta, ...
seng.Iip (or senglip)
pang, tong-an, congi,
cwungang, thongil
ilk.ko, ilk.hinta, ilk.e
ilk.nun (or ilknun), malk.ta (or malkta),
pulk.ci (or pulkci), ...
A Reference Grammar of Korean
PART I 11
Hankul Initial
Example Final
Examples
aw
-
lp. + p h 0
nelp.e, nelp.hinta
lp(.)
nelp.ko (or nelpko), nelp.ci (or nelpci),
nelp.ni (or nelpni)
53
—
lph. + 0
ulph.e
lph(-)
ulph.ko (or ulphko),
ulph.nun (or ulphnun), ...
an
-
1m. + m 0
talm.un
lm(.)
talm.nun (or talmnun)
(But Cf §8.1.
1 talm.ko, talm.ta, talm.ci.)
5E
-
1th. + 0
halth.e, halth.un
lth(.)
halth.ko (or halthko),
halth.nun (or halthnun)
a*
-
lh. + 0
ilh.e, ilh.un
lh(.)
ilh.ko (or ilhko), ilh.nun (or ilhnun), ...
a/.
-
Is + space
kols, tols, ols
UA
—
ps + space
kaps
ps. + s 0
eps. s up.nita
eps.e
ps(.)
eps.ta (or epsta), eps.nun (or epsnun), ...
LA
—
nc. 4- c 0
anc.ci
anc.e
nc(.)
anc.ko (or ancko), anc.nun (or ancnun), ...
us
-
nh. + 0
manh.i
nh(.)
manh.ta (or manhta), manh.ni (or manhni)
MEDIALS
Hankul
Romanized
Examples
1
i
pi, Is.ta, oi
A
wi
twi, chwuwi, wiseng
A
uy
uyca, cwuuy, uyuy, huyta
T
u after p ph pp m
pul, phul, pun, mun
wu elsewhere
wuli, hwu, nwun, twul, kkwum
7T
.yu after a e o (u)
ye.yu, ca.yu, so.yu
yu elsewhere
yuli, kyul, hyung, wuyu
—
u
un, kum, khuta; papputa (§1.7)
H
e
emeni, khe (yo), pappe, tewuk
we
wenca, kwen, il-wel
H
.ye after a e o (u)
ca.yen, he.yeng, mo.ye
HI
ye elsewhere
yek, n yen, ^en.ay, kyewul
ey
ney, cey-il, kakey, pheyn
Hi
.yey after a e o (u)
a.yey, no.yey = no'yey, ku.yey
yey elsewhere
yeysan, ‘yeypay, kyeysita
Hi
.wey after a e o u
[rare]
wey elsewhere
weynq il, kwey
_L
0
oi, hao, mom, ton, os, kopta
II
.yo after a e o (u wu)
hwa.yo(il), he.yong, so.yong
yo elsewhere
yokwu, yongpi, iyo, phyo
12 PART I
A Reference Grammar of Korean
Hankul Romanized Examples
4 oy oykwuk, kkoy, sloy, poynta
1 - a ama, koa, tong-an, hayan
j- 1 - wa wa (se), pwa, towa, cwasek
1 = .ya after a e o (u) a.ya, se.yang, co.yak
ya elsewhere yacwung, kyawus, iya
H ay ayki, kkay, hayan, sensayng
H .yay after a e o (u wu) ha.yay, he.yay, ppo.yay, ppu.yay
yay elsewhere yayki (= iyaki)
41 way way, kkway, insway, twayci
The Hankul symbol for we can be written A, but A is preferred.
1.5. Reinforcement (- q ).
The rules for the treatment of what we call “reinforcement” are somewhat complicated. Roughly
speaking, the linguistic facts are as follows. When certain words or parts of words are attached to
others that begin with p t c k or s, that consonant is doubled to pp, tt, cc, kk, ss. When they are
attached to certain other elements that begin with “zero” (a vowel) or h, the vowel or h is preceded by
t; when they are attached to certain elements that begin with m or n, those onsets are preceded by
syllable-final n; when they are attached to certain elements that begin with y or i, the pronunciation is
l-ny-l or /—ni—/ after a consonant and l-nny-l or /••• nni-/ after a vowel, but -lq y™ and -lq i-
are realized as /— lly —/ and /—Hi—/. Reinforcement of — n i — *-nq i— /—nni—/ and — n y— *-nq y—
l-nny—l, sometimes called “n-epenthesis” (not to be confused with the nasal epenthesis of §2.10.4) is
so pervasive that it passes largely unnoticed: musunq iwus ‘what neighbor’, onq yatang ‘the entire
opposition party’, sinq yetang ‘a new party in power’, sinhonq 'yehayng ‘honeymoon trip’. But such
reinforcement does not occur before the noun i ‘person; fact/act; ... ’, the particle iya/ya, the particle
(iyo)/yo, or the copula stem i-. The peculiarities of the phoneme string /yey/ (§4.3) are reflected in the
pronunciations /silleysan/ for silq yeysan ‘real budget’, /kalleycengita/ for kalq yeyceng ita, and
/musunneyki/ for musunq yeyki («- yayki < iyaki) ‘what talk’. As a result of vowel raising tonq
(iyaki > yayki >) yeyki -* /tonneyki/ ‘money talk’ may sound just like tonq (nayki >) neyki
‘gambling’. When the juncture after the accusative particle is dropped, os ul (I) ip.e ‘wears a garment’
is usually said as /osulipe/ with a flap [r] but you may sometimes instead hear /osullipe/. When the
particle itself drops, os [ul] ip.e, you usually hear /onnipe/ = osq ip.e rather than /otipe/ = os ip.e
with [d]. When the particle gets dropped in mun (ul) yel.e ‘opens the door’ the phrase is usually said
as munq yel.e, pronounced /munnyele/ and not mu-nye-le. There are a few lexical exceptions to the
n-epenthesis rule, notably Vuk-i [yugi] rather than (?) 1 yukq-i [yungni] ‘six-two; sixty-two’, familiar in
the term ‘yuk-i-o ‘6-25’ (= ~ sapyen/tong. Ian ‘the North Korean invasion of June 25th, 1950’)
which is pronounced /yukio/ or contracted to /yukyo/. We list seven types of examples:
(1) Native Korean sequences (usually compounds of noun + noun) in which the first element ends
in a vowel.
(2) Native Korean compounds in which the first element ends in a consonant.
(3) Chinese loanwords (originally, in Chinese, compounds - usually of two syllables) in which the
first element ends in a vowel.
(4) Chinese loanwords in which the first element ends in a sonant (m n ng 1).
(5) Chinese loanwords in which the first element ends in 1 and the second begins with t, c, or s. If
both elements are bound, the reinforcement is obligatory; if one element is free, the reinforcement is
usual but there are exceptions.
(6) Korean verb forms consisting of a stem that ends in a nasal norm (from a linguistic point of
view also those that end in nc, lm, lp, Iph, 1 th, and for some speakers lk) plus an attached ending that
begins with t, c, 1, or s. Before s the reductions of final nh and Ih can be added.
(7) A sequence of the prospective modifier (-ul) followed by a noun that begins with p, t, c, k, or s.
A Reference Grammar of Korean
PART I 13
Spellings in both the North and the South ignore the reinforcement that occurs automatically in
Type 5 (Chinese loanwords with 1 + s, 1 + t, 1 + c). In the Yale Romanization we prefer to write lqs,
Iqt, and lqc for these words since in the non-Chinese words there are contrasts with unreinforced Is, It,
and lc. Moreover, it is not always easy to know that a particular word is of Chinese origin.
Both of the Hankul spelling systems also ignore the reinforcement that occurs automatically in
Type 6 (verb forms), and indeed there are dialects which do not reinforce these forms (see §8.2.1).
Instead of here writing -q, as one might prefer, the Yale Romanization uses a dot as a reminder
notation: -m.t-, -n.c-, ... .
Type 7 is ignored by the South Korean spelling but sometimes indicated in the north, and it was
often written in early Hankul texts with a “glottal stop” symbol corresponding to the -q with which we
mark it. (The symbol is like the Hankul h but without the short stroke at the top: o )
Types 3 and 4 (the other Chinese loanwords) are treated the same as Types 1 and 2 (the Korean
words) by the North Korean system, but in practice there is variation, perhaps owing to an indecision
over whether to admit some of the reinforced forms as “standard”. The NK dictionary (NKd) often
omits the apostrophe in the entry spelling for words of these types, and gives the pronunciation
separately: “saken [kken]”, “sanpo [ppo]”. Yet for other entries both the apostrophe and the separate
pronunciation are given: “ka’pep [ppep]”, “mun’ca [cca]”. The South Korean system, too, sometimes
treats Types 3 and 4 as if they were 1 and 2 respectively, but many people are inconsistent or forgetful
and ignore the reinforcement in the Chinese loanwords. 'Yi Ungpayk (458) advises us to write the
postvocalic “-s” for Type 3 only when there is a minimal contrast with another word that does not
have the reinforcement: kaqpep ‘addition’, kapep [lit] ‘family tradition’; seqca [obsolescent?] ‘letter
(epistle; character)’, seca ‘illegitimate child’; Hqkwa ‘science’, i-kwa ‘lesson two’; hoq-swu ‘number
of households’, hoswu ‘lake’; choqcem ‘focal point’, cho-cem ‘vinegar shop’ (? - not found in
dictionaries); yoqcem ‘main point’, yo-cem ‘mattress shop’ (? - not found in dictionaries).
For Types 1 and 2, the earlier North Korean rule is very simple: at the end of the prior element
add an apostrophe, called sai phyo ‘between-mark’ and looking much like the left side of the Hankul
letter A. The Yale Romanization uses a similar device: add q at the end of the prior element. (We
prefer to avoid the apostrophe because it is useful for other purposes, such as showing abbreviations.)
In the South Korean system a final s is added to the prior syllable when it ends in a vowel (Type 1),
and the reinforcement is ignored when the prior syllable ends in a consonant (Type 2), except that as
the later element the morpheme i ‘tooth’ is spelled ni = /-(n) ni/ (2). At one time, following the
practice of the early texts, the -s— was written as a separate syllable all by itself and called sai sios
‘in-between s’; at that time (the early 1930s) it was used for both Type 1 and Type 2.
The following table shows examples of all seven types with the different treatments, together with
the pronunciation and the phonemic shape as transcribed in Yale Romanization symbols, but the
reduction of syllable-excess at the end of nouns is not shown. Space and hyphen are retained from the
notation on the left and do not necessarily reflect the practice for the particular phrase in either the
south or the north, since compounds and short phrases are usually written without a break in both parts
of the country.
TABLE OF REINFORCEMENT TYPES
Romanization
South Korea
North Korea
Pronunciation
1. twlq path
twis path
twi’ path
pay’ nolay
/twlppath/
payq nolay
pays nolay
/paynnolay/
peykayq Is
peykays is
peykay’ is
/peykaynnis/
ku-kkaciq il
ku-kkacis il
ku-kkacis’ il
/kukkacinnil/
2. wiq iq-mom
wis is-mom
[*wi’ i’-mom]
/winnimmom/ 1
mulq kyel
mul kyel
mul’ kyel
kang’ ka
cip(’) il
/mulkkyel/
kangq ka
kang ka
/kangkka/
cip(q) il
cip il
/cipil, cimnil/
14 PART I
A Reference Grammar of Korean
petulq iph
petul iph
petul’ iph
kas’ yang(thay)
mul’ yak
keth’ iph
yeys’ iyaki
kath’ yakta
kyelmak’ yem
songkos’ i
/petulliph/
kas yang(thay)
kas yang(thay)
/kannyang(thay)/
mulq yak
mul yak
/mullyak/
kethq iph
keth iph
/kenniph/
yeysq iyaki
yeys iyaki
/yeynniyaki/
kethq yakta
keth yakta
/kennyaktta/
kyelmakq yem 1 2
kyelmak yem
/kyelmangnyem/
songkosq (n)i
songkos ni
/songkonni/
tesq (n)i
tes ni
tes’ i
/tenni/
okq-(n)i
ok-ni
ok’-i
/ongni/
petq-(n)i
pet-ni
pet’-i
/penni/
ppetulengq (n)i
ppetuleng ni
ppetuleng’ i
ka’pep
/ppetulengni/
3.
kaqpep
kaspep
/kappep/
saqken
sasken
sa(’)ken, saken
/sakken/
4.
munqca 3
munca
mun’ca
/muncca/
sanqpo
sanpo
san’po, sanpo
/samppo / 4
inqki
inki
in’ki, inki
/ingkki / 5
5.
cengqyel
cengyel
cengyel
/cengnyel/
kyelqsan
kyelsan
kyelsan
/kyelssan/
kyelqtan
kyeltan
kyeltan
/kyelttan/
6 .
kyelqceng
kyelceng
kyelceng
/kyelcceng/
slm.ta, slm.ko
simta, simko
simta, simko
/slmtta/, /slmkko/
7.
celm.ta, celm.ko
celmta, celmko
celmta, celmko
/cemtta/, /cemkko/
halq kes
hal kes
hal’ kes, hal kes
/halkkes/
See also §4.2 (end); silh(q)-cung in Part II.
1 *- /winninmom/; Cf CM 1:99. For all uses of the morpheme meaning ‘up’ NK
standardizes wu; the adnoun (or prefix) should be wuq but they have standardized it as
wus, the spelling that is used in the south for those phrases that have preserved the older
vowel nucleus. See §5.3. This word is not carried by NKd.
2 Though the parts are Chinese, this second-degree compound is here treated as if Korean.
3 = kulqca ‘letters, written characters’.
4 *- /sanppo/
5 *- /inkki/
For reasons not clearly understood a fair number of Chinese loanmorphs show a marked tendency
to induce reinforcement as the final member of a compound: -(q)kwa ‘course’ or ‘section’, -(q)kwen
‘chit’ or ‘sphere’ or ‘privilege’, -(q)ken ‘case, matter’ (Cf saqken, coqken, anqken), -(q)ka ‘price’
(CF tayqka, yuqka, ^uqka, 'yemqka), -(q)kwi ‘couplet’ (Cf eqkwi, ^enqkwi), -(q)ki ‘feeling’ (Cf
yunq-ki), -(q)kyek ‘standing, rule, grammatical case’ (Cf cwuq-kyek, inqkyek), -(q)ca ‘written
character’, -(q)cem ‘point’, -(q)cang ‘document’, -(q)coy ‘crime’, -(q)cung ‘illness’, -(q)pyeng
‘illness’, -(q)pep ‘law, rule’, -(q)po ‘step’ (sanqpo); -(q)swu ‘number’ (Cf chiq-swu ‘size, measure’
£ chi-swu ‘number of inches’); ... . See also -(q)seng ‘-ness’ and cek ‘-ic’ in Part II. The syntactic
relationship of the two morphemes in Chinese is irrelevant to the reinforcement: silqkwen can mean
either ‘real power’ or ‘lose power’; the verbal noun sanqpo ‘stroll’ comes from a Chinese verb-object
phrase (‘scatter one’s steps’).
For certain phrases we find vacillation in whether to reinforce or not: i kkoch ilum /ikkotilum/
‘the name of this flower’ and ku umsik ilum /kuumsikilum/ ‘the name of that food’ (both M 1:1:390)
are more commonly said as i kkochq ilum /ikkonnilum/ and ku umsikq ilum /kuumsingnilum/.
Similarly san kkochq ilum ‘names of mountain flowers’, sothq ilum ‘names of pots’, (yang)kokq
ilum ‘names of songs’, *yekaykq ilum ‘passenger names’. Yet tayhak ilum /tayhakilum/ is preferred
A Reference Grammar of Korean
PART I 15
to tayhakq ilum /tayhangnilum/ ‘college name’. (On -q y- and -q i- see §4.4.) The word for
‘carsick(ness)’ is usually said as chaq melmi /chammelmi/ but dictionaries ignore the reinforcement;
similarly ignored is the usual reinforcement of chiq sol ‘tooth brush’ and chiq-swu ‘measure’. Certain
phrases, however, are always reinforced: tankolq tapang ‘a favorite teashop’, tankolq son nim ‘a
regular customer’ (M 1:1:390); petulq kaci = petu’ namuq kaci ‘a willow(tree) branch’ (the
apostrophe marks an elision); ... . The phrase chinkwu(q) cip ‘a friend’s house’ can rime with iwus cip
‘neighboring house’ but wuli cip ‘our house’ is never *wuliq cip. And others are never reinforced: tol
path ‘a field of stones’ (not *tolq path). NKd indicates reinforcement in chongka ‘gun mount’ but no
other source confirms that, so it may be a mistake. NKd lacks an indication of reinforcement for
cwungqcung ‘grave illness’ and kyengqcung ‘light illness’ (as found in LHS and in Kim Minswu and
Hong Wungsen), and no dictionary indicates reinforcement in hwa( ? q)cung ‘a flareup of anger’.
1.6. Initial 1 and n .
Words beginning with a morpheme which has a basic and/or historical shape that begins with 1 are
pronounced with an initial n. Such words are written in South Korea with an initial n but in North
Korea they are written with an initial 1; in this book we write 'n- for these words. But before i and y
neither n nor 1 is pronounced. The South Koreans follow the pronunciation and begin the word y™ or
i-, but the North Koreans write ly~ and li- or ny- and ni- depending on the basic and/or historical
shape of the morpheme; and here we write 'y— and M- or n y™ and n i™, with a few cases of “(n)”
owing to the inconsistencies found in spelling practices. The following examples show the differences:
Romanization
South Korea
North Korea
Pronunciation
Meaning
"yenkam
yenkam
nyenkam
/yenkam/
‘yearbook’
"ik.myeng
ikmyeng
nikmyeng
/ingmyeng/
‘anonymity’
n itho
itho
nitho
/itho/
‘mud’
'yensup
yensup
lyensup
/yensup/
‘drill (study)’
*yey
yey
lyey
/yey/
‘ceremony’
'ihay
ihay
lihay
/Ihay/
‘understanding’
'nonmun
nonmun
lonmun
/nonmun/
‘treatise’
'naywel
naywel
laywel
/naywel/
‘next month’
The family name 'Yi, in English variously spelled Lee, Li, Yi, Ree, Ri, Rhee, ... , is particularly
troublesome. In South Korea it is spelled I and in North Korea Li. The Yale Romanization should be 'I
but that looks awkward, so we make an arbitrary exception and write 'Yi, preserving a resemblance to
the form that is familiar from references to the “Yi dynasty”. (The phoneme III can be deemed to carry
a nondistinctive initial y- which we ignore except for this name.) The pronunciation is i, but ni is (or
was) used by northerners, though the Phyengyang authorities have promoted saying /li/ with an initial
flap, like that used in recent foreign loanwords. For them, history has reversed itself: /li/ -* /ni/ -* I'll
-* /li/. Even in South Korea ‘Miss 'Yi’ is always Misu Li [ri] and never *Misu 'Yi [i]. Other proper
names beginning with (such as 'im, 'Yang, and even *No) may be given reading pronunciations
with the flap, and these could become the spoken norm for that part of the country. The authenticity of
the pronunciation ni is quite clear from attestations such as 1881 Ridel 20, who has that version both
in Hankul and in his transcription, and (23) writes “Ni ryengkam” for ‘the venerable 'Yi’. 1887 Scott
writes “ni-” for the etymological ni- in a fair number of words, and also has a few examples that go
back to li-, such as ninsoyk (123) < linsoyk ‘avaricious’; he writes ryemnye ‘fear’ in one place (127)
yet nyemnye ‘anxiety’ in another (176). At the same time, he says (149n) “Though spelt rika it is read
ika” with reference to 'l ka ‘profit’ [nominative] (Cf 166). The pronunciation of li as ni is reported in
the first part of the sixteenth century: the surname 'YI < ~Ll is attested as "ni in ? 1517- Pak 1:3a and
m 'chyen ‘profit money’ < "lichyen occurs in the same work (1:34a) and also in ? 1517- 'No 1:13a,
though in the latter the word appears again (2:60a) as li'chyen with the l intact. The word nyenskon
(1527 Cahoy l:8a= 14b) ‘lotus root’ was lexicalized from the phrase LYEN s KON ‘root of the lotus’.
16 PART I
A Reference Grammar of Korean
All of the words involved in this section are of Chinese origin. Other words that were spelled with
ny- and ni™ in Middle Korean are now written y- and i— , as pronounced, but when the nasal persists
in compounds the spelling in South Korea follows that pronunciation: songkosq (n)i ‘cuspid’ is written
as “songkos ni” in the south but “songkos* i” in the north (without our generous spaces, of course).
There are inconsistencies in the decisions, as exemplified by the SK treatment of certain words in
group 2, for when the nasal is pronounced in (n)i ‘tooth’ it is favored by an attention neglected in the
case of iph ‘leaf’, II ‘work, event’, and iyaki ‘tale’.
The word for ‘glass’ is historically lyuli, but North Korea spells it yuli (as we expect of South
Korea). That means that the phrase /sayngnyuli/ ‘stained glass’ must be treated as saykq yuli, not
sayk 'yuli or sayk-lyuli. A similar case: ssang-pongq yaktay /ssangpongnyakttay/ ‘Bactrian camel’,
historically lyaktay.
The following words begin with an etymological I- but are written phonemically with n- in
North Korea as well as South Korea: nasa ‘screw, spiral’ (Cf 'nasa ‘woolen cloth’), naphal ‘trumpet’,
nok ‘rust’ (Cf 'nok ‘stipend’), nampho(-tung) ‘lamp’, no ‘oar’, nwu ‘loft, pavilion’, nwuki
‘dampness’, nwu-nwu-i ‘frequently’. Cf Mkk 1960:3:25. The morpheme 'yen is spelled lyen by North
Koreans in 'yenq-ie ‘consecutively’ but yen in yen-kephe = yen-kephu ‘successively’ and in yen-hay
and yenpang ‘continuously’ (a point missed by KEd). Cf Mkk 1960:4:23.
The word /silyen/ ‘disappointment in love’ is etymologically sil-lyen; it is spelled sil.yen in the
south and sillyen in the north, so we will write it sil'yen. A similar case is sal'yuk ‘massacre’,
pronounced /sallyuk/. We would expect to write no'yey for /no(.y)ey/ ‘slave’ (Cf halyey ‘male slave’)
but the spelling in both north and south is no.yey. The word olyu ‘mistake’ (< o-lyu) has a common
variant pronunciation without the 1 and that is standardized in NKd with the spelling o.yu.
An epenthetic (intrusive) -1- is found before lyl in a few compounds: ceylyem < cey-yem
‘manufacturing salt’, holyem < ho-yem ‘Chinese rough salt’; ph y eylyem < phyey-yem ‘pneumonia’
(spelled phyeylyem in the south but pheyyem in the north so we will write it ph y ey'yem), thoylyem
< thoy-yem ‘pouring hot broth over rice or noodles a little at a time to heat them up’ - and perhaps
(yang-)hwalyo ‘carpet’ and polyo ‘large fancy cushion’, if these are properly derived from yo
‘mattress’ < zywoh < *nhyok, a Chinese import that underwent early naturalization.
In a few words an etymological n is commonly pronounced as III: taylo < tay-no ‘great anger’,
huylo < huy-no ‘joy and anger’, elwul < e-nwul ‘inarticulate’, yalyo < ya-nyo ‘annoyance,
interruption’.
A small group of morphemes have the basic shapes lyul and lyel. These morphemes follow a
special pronunciation rule. After a vowel or n, the 1 unexpectedly drops. (The 1 actually surfaces only
after some consonant other than n, including 1 itself, and then it takes the reflex Ini so that we would
have no idea these morphemes begin with a basic 1 rather than n without additional information from
the history or from dialect pronunciations.) The alternation is ignored in the northern spelling but
phonemically noted in the south, so we mark it in our Romanization with -*y-. Examples will be found
in §4.8. When these morphemes are attached to a morpheme that ends with n we insert a dot (cin.'yel
‘exhibition’, cen.'yul ‘trembling’) to remind us of the morpheme boundary in the SK spelling
(“cln.yel”, “cen.yul”) as contrasted with the NK spelling (“cln.lyel” = /clnyel/, “cen.lyul” =
/cenyul/). When the preceding morpheme ends with i, wu, yu, or a digraph of vowel + y, there is no
need for the dot to mark the boundary, so we write pl'yel ‘(being) nasty’, pi'yul ‘ratio’, swu'yel
‘numerical progression’, kyu'yul ‘rules’. But after the other vowels, the SK spelling requires us to
insert a dot to prevent interpreting the string of vowel + y as a digraph: to.'yel ‘lining up’ (NK
“tolyel”, SK “to.yel”), pha.'yel ‘explosion’ (NK “phalyel”, SK “pha.yel”). KEd was inconsistent in
not writing the dot for those cases. On problems of spelling and interpreting words containing the
morpheme lyo ‘fee’, see the entry in Part II. See also §4.4.
In the non-Chinese vocabulary the initial nasal of words beginning nye or ni was eventually
dropped, so that today ‘tooth’ is pronounced i in both the north and the south. It is difficult to assign
A Reference Grammar of Korean
PART I 17
this change to a particular time. It may have set in first for verb stems: we find yey- in ? 1660-
Kyeychwuk and (with raised vowel) i- in 1876 Kakok for what was earlier "nyey- and still earlier °nye-
and °ni- ‘go’. In Middle Korean the prevalent form for ‘put in’ was nyeh- > dialect yeh-, but a
dispalatalized neh- is attested in 1466 Kup (see p.47). Modern ic- ‘forget’ for earlier nic- is attested in
the middle of the 18th century and perhaps as early as 1660. There may have been dialect doublets
with ny- competing with y- and dispalatalized as n- from fairly early; the ancestor of modern yeki-
‘regard (as)’, first so attested in ? 1800 Hancwung, was nye'ki- in 1481 Samkang and Twusi, but the
dispalatalized version ne'ki- was prevalent from 1449 (Kok) right on down (including 1481 Twusi).
The adjective nyeth- ‘shallow’ is spelled yeth- in 7 1517- Pak-cho and later, and yath- in 1608
Twuchang cip.yo. On the other hand, there is but a single example of net- (in 1763 Haytong) for “yel¬
ite ancestor of modern ye-I- ‘open’, and there are no reports of an initial nasal in modern dialects, so
that the one example is suspect. In most words that began with ni~ there is no evidence that the n
dropped until quite late: nik- > ik- ‘ripen’, "niz- > I(s)- ‘continue’, nilk- > ilk- ‘read’, ni > i
‘tooth’ or ‘flea’, niph > iph ‘leaf’, ni'mah > ima ‘forehead’, ~nimca[h] > Imca ‘owner’, nil'kwup
Inil 'kwop > ilkop ‘seven’.For many of these words, some of the northern dialects have retained
the nasal: Kim Thaykyun cites ni ‘tooth’ and ‘flea’, niph ‘leaf’, nilkop ‘seven’, nlmca ‘owner’, nilk-
‘read’, nip(-hi)- ‘(cause to) wear’, nl(s)- ‘continue’, nyeth-/nyath- ‘shallow’. The earliest texts had
both "nil- ‘arise’ and ~il- ‘come into being’; could they have once been etymologically the same? The
only modern words pronounced with initial ny- seem to be nyesek ‘rascal (of a man/boy)’ and nyen
‘bitch (of a woman)’, of somewhat obscure etymology but both probably involving the Chinese
morpheme nye ‘woman’. The modern yeph ‘(be)side’ was nyeph in 1617 Sin-sok (*Yel 3:24) and just
nyep in earlier texts (nye p u[']lwo 1459 Wei 2:36a, nye p i'la 1459 Wei 2:17b), but dispalatalized
nep.hu.lwo appears in ? 1775 Han-Cheng 204d. It should be kept in mind that Korean also has words
beginning i- and y- which have never had variant versions with a nasal initial, such as ip < ip
‘mouth’, isul < i'sul ‘dew’, ilang < i'lang ‘paddy ridge’, ili < ‘Uhl ‘wolf’, iki- < i'ki-H'kuy-
‘win’, ilh- < ilh- ‘lose’; yewi- < ye'wuy- ‘get thin’, yetelp < ye'tulp ‘eight’, yeses < ye'sus ‘six’.
These words have not been reported with a nasal initial in dialects, with the exception of Hoylyeng
nyessay ‘six days’ (Kim Thaykyun 1986:380a), and that form is surprising, if correct, in view of MK
ye'ssway (1462 'Nung 6:17a) and ye'sway (s pa'm oy 1481 Twusi 10:4a). But nasal versions are
reported for the modern words (yeyki < yayki <) iyaki ‘story, tale’, which was first attested in 1775
as niyaki, and ieng ‘(roofing) thatch’, first attested in the 1730 text of Chengkwu yengen as niyeng.
(Also reported in 20th-century materials: nikki for ikki < isk ‘moss’, nisak for isak < isak ‘ear of
grain’ and nyemul for yemul < ye 'mulq ‘to open’; see King 199lb:6.) The nasal dispalatalization in
the north and the palatal denasalization in the south is part of a more general process of reducing initial
strings of apical (dental or alveolar) + palatal, as described in §4.4.
In Hamkyeng the Chinese morphemes that begin with a basic y- or i- (and uy, which merges
with HI even initially in these dialects) are often treated if they were from ny- or ni- when attached to
a Chinese morpheme that ends in a consonant, even if both morphemes are bound. (Choy Iceng 1960.)
We can note these pronunciations with the morphophonemic —q:
Standard
Hamkyeng
Pronounced
Meaning
mok.yoil
mokq.yoil
/mongnyoil/
‘Thursday’
sap.ip
sapq.ip
/samnip/
‘insertion’
cen.ya
cenq.ya
/cennya/
‘the night before’
tham.yok
thamq.yok
/thamnyok/
‘greed’
cung.ye
cungq.ye
/cungnye/
‘donation’
cel.yak
celq.yak
/cellyak/
‘economizing’
Such pronunciations are not new. 1889 Imbault-Huart 66 has sik.nyem for sik.yem ‘salt’, which must
have been pronounced /singnyem/= sikq.yem.
18 PART I
A Reference Grammar of Korean
1.7. Hankul spelling of u after labials.
The back unrounded high vowel u does not occur after a labial (p ph pp m). When expected, it is
replaced by the rounded vowel wu. That is why the Yale Romanization writes wu as just u after a
labial. Our notation of pu can be regarded as a convenient abbreviation of pwu. But that is true only
when we are speaking of modern Korean. Until sometime after 1700 Korean distinguished labial-onset
syllables with the unrounded vowel from their counterparts with the rounded vowel, so that when we
cite forms from that period (or specific Hankul spellings from even later) we must distinguish pu from
pwu and mu! from mwul, just as we keep distinct the zero-onset syllables that begin with u and wu.
The present-day Hankul spelling systems usually write wu after a labial within a morpheme (mun is
written m + wu + n, pun is p + wu + n, phun is ph + wu + n, and ppun is pp + wu + n), but
choose between basic wu and u after the zero onset of a morpheme occurring after an element that
ends in a labial: kem.un ••• ‘ ••• that is black’ and kum-un ‘gold and silver’, but kam.wu ‘a welcome
rain’. We know whether to write wu or u by recalling other words that contain the same morphemes in
environments where the two vowels are distinguished: cak.un — ‘ - that is small’ and un ‘silver’, but
wusan ‘umbrella’.
There are exceptions. Stems that end in the high back vowel after a labial are given the Hankul
equivalent of our Romanized abbreviation [w]u, so that the second syllable of kippu- ‘be happy’ is
spelled pp + u even though it is pronounced with a rounded vowel, and the second syllable of kophu-
‘be empty, hungry’ is spelled ph + u. Morphophonemically, the spelled “u” behaves like any other
case of u, in that it drops when the infinitive ending is attached: kippe, kopha. We also see u
sometimes in derived forms: kwucepun hata ‘is untidy’ may be spelled with Hankul p + u + n,
reflecting its derivation from kwucep (sulepta) + -un. But lexical derivations of that sort are largely
ignored under the current spelling standards of both parts of Korea, where the word is written with p
+ wu + n, as it sounds. The spelling with u is used in kumum (nal) ‘last day of the month’ because
kumum is an etymological variant of kem.um ‘being black’ (i.e. ‘dark of the moon’); compare polum
‘middle day of the month’, an etymological variant of polk- + -um = palk.um ‘being bright’. And
Hankul regularly spells u for the “inserted vowel” used to help pronounce certain final consonants and
clusters in foreign words, so that the second syllable of Aphulikha ‘Africa’ is written ph + u rather
than ph + wu.
Despite the Hankul spelling, na ppun ‘only me’ and nappun ••• ‘bad’ sound the same, and the
three expressions ku mun ‘that door’, kum un ‘as for gold’, and kum-un ‘gold and silver’ are
homonyms. The lips are rounded (and usually protruded) throughout the last syllable of each of these
phrases.
1.8. Word division and internal punctuation.
As in English and many other languages, people vary considerably when deciding questions of
spacing and punctuation. One word or two words? Hyphenate or run together? Until the 20th century
Hankul texts were commonly written with no space between words, as is the traditional practice in
Chinese and Japanese. The modern use of word spaces began with Toklip sinmun 1 (1896) and was
continued by Cwu Sikyeng (1907 “Kwuk.e wa kwuk.mun uy phil.yo”). A similar usage in Japan,
found in hiragana newspapers of the Meiji period and in elementary textbooks from 1884, never took
hold. (There are examples of word spacing in manuscripts of 1272 left by Shinran.) The Korean
decision to use spacing may have been independent, or it may have been influenced by the
orthographies of European languages. In 1897 'Yi Pongmun (Kwukmun cengli) marked the ends of
words by putting little circles below and slightly to the right in the vertical line of Hankul.
In writing Korean the most common practice is to run together things that are spoken together in
phrases, especially noun + particle (+ particle) and verb + particle (+ particle). People disagree on
whether to write sequences such as adnoun (prenoun) + noun and noun + noun (+ noun + ... ) as
separate words. Writers like to attach very short elements, especially those of one syllable, to longer
elements that are contiguous to them. That fact, combined with hazy notions of grammar to begin with,
A Reference Grammar of Korean
PART I 19
accounts for the writing of particles together with the preceding noun or verb. Considerable indecision
prevails in treating noun compounds, numeral + counter, adverb + verb, and those verb compounds
in which the prior element is in a “free” form, carrying an ending, such as the infinitive, rather than
just the stem alone. If the form becomes very long, people are apt to break it with a space; if the whole
sequence is relatively short, it will likely be run together. A sequence of verb modifier + noun poses a
problem, for the modifier is always the head of the prior construction, which can be quite long, but in
pronunciation a pause is more likely to come before rather than after the modifier. Unmarked object +
verb are often run together, especially if the object is short and the expression common, like pap mek-
‘eat food’.
To a certain extent, spacing decisions reflect the degrees of potential pause between various
constructions, and we can summarize that in a table of linkage (juncture in the rough sense of the
word), from openmost to tightest, with examples:
1. Topic (noun + un/nun) + verb
2. Subject (noun + i/ka) + verb
3. Indirect object (noun + hanthey
or eykey) + verb
4. Object (noun + ul/lul) + verb
5. Unmarked subject + verb
6. Unmarked indirect object + verb
7. Unmarked object + verb
8. Adverb + verb
9. Noun + noun
10. Adnoun + noun
11. Adnoun + quasi-free noun
12. Modifier + noun
13. Modifier 4- quasi-free noun
14. Numeral + noun
15. Numeral -I- counter
16. Verb (infinitive/gerund) + aux verb
17. Noun + copula
18. Quasi-free noun + copula
19. Noun + particle
20. Verb + particle
21. Particle + particle (+ particle)
22. Verb stem + ending
ayki nun mek.e; pap un mek.e
ayki ka mek.e
ayki hanthey cwue
pap ul mek.e
ayki mek.e
ayki cwue
(pipimq) pap mek.e
manh.i mek.e; cal mek.e; mos mek.e
hak.kyo sensayng; kkoch path
ku salam, ku ay
ku kes
kulen salam; halq salam
kulen kes; halq kes
twu son
twu mali
mek.e twue; mek.ko siph.ta
pap ita
- kes ita
ayki nun; pap ul; cip i
mek.e yo; mek.e to; mek.ko nun
• ey se; ••• ey se to
mek-e; khu-n; kitali-ci;
ka-syess.ess.keyss. 3 up.nikka
This list does not, of course, exhaust the possibilities. Places in the table should probably be
found for other constructions. Where real anxiety arises, for the native speaker and the linguist alike, is
in conflicting combinations of linkage constructions, as in Haksayng hanthey phyenci sse cwulq kes
iey yo ‘(He) will probably write the letter for the student’, which has the following constituency:
Haksayng hanthey phyenci sse cwulq kes iey yo
NOUN PARTICLE NOUN VERB AUX NOUN STEM+ENDING PARTICLE
nominal - adnominal adverbial - verbal
adverbial-verbal
adnominal-
nominal-
verbal —
nominal
-copula (inf)
particle
(adverbial) sentence
20 PART I
A Reference Grammar of Korean
But the phonological bonds, the words most likely to run together in pronunciation, would be in the
following order of closeknitness: (1) iey yo; (2) haksayng hanthey; (3) kes ley; (4) sse cwul; (5)
cwulq kes; (6) phyenci sse. This means that if we were to make only one pause it would likely be
Haksayng hanthey I phyenci sse cwulq kes iey yo, if we made two pauses it would be Haksayng
hanthey I phyenci 1 sse cwulq kes iey yo, and if we should make a third pause (unnatural for this
sentence) it would be Haksayng hanthey I phyenci I sse cwul I kes iey yo. If we were to make four
pauses (still more unnatural) the sentence would be Haksayng hanthey I phyenci I sse I cwul I kes
iey yo. Any further pauses, as in taking the next step kes I iey yo and the final steps emeni I hanthey
and ie(y) I yo, would be artificial.
From this example it is easy to see the all too frequent conflict between the phonological bondage,
or closeknitness of pronunciation, and the immediate constituency, or closeness of grammatical ties.
Similar problems in English are exemplified by such expressions as “the highest scoring team”, “sharp
bladed”, “three hundred and first”, “your nearest store”.
In Hankul texts prepared under my supervision, such as the examples in KEd, we have been
liberal in word division. For types 1 through 16 above, either space or hyphen is used (hyphen for
some of the very frequent or very short combinations of types 7 through 16); for types 17 through 21a
dot is inserted to set a particle off from the preceding word to which it is attached; and only examples
of type 22 are regularly spelled with no internal punctuation. Some such rules, perhaps, could
eventually be incorporated into normal Hankul writing, at least in school textbooks, where it is
important to reduce ambiguity to a minimum. (Notice the unusually generous word division in the
North Korean journal Mkk 1961:4:37-40.) Quite a few homophonous phrases can be distinguished by
inserting a pause or “open juncture”:
/hoysaey( )nakanta/ = hoysa ey na-kanta ‘goes off to the office’
/hoysaeyna( )kanta/ = hoysa ey ’na kanta ‘goes off to the office or the like’
/na( )kakiceney/ = na kaki cen ey ‘before I go’
/nakakiceney/ = na-kaki cen ey ‘before going out’
/wuliyekwaney( )kaca/ = wuli *yekwan ey kaca ‘let’s go to our inn’
/wuli yekwaneykaca/ = wuli 'yekwan ey kaca ‘let’s go to an inn’
/wuli yekwaneyka ca/ = wuli ‘yekwan ey ka ca ‘let’s go to an inn to sleep’
/kuchaykina omyen( )cokheytta/ = ku chayk ina omyen coh.keyss.ta ‘I hope that the book comes,
or something’
/kuchayki naomyen( )cokheytta/ = ku chayk i na-omyen coh.keyss.ta ‘I hope the book comes
out’
The following two expressions will both translate as ‘the rice tastes good’ yet they represent different
constructions:
/pap masi( )cotha/ = pap (un) mas i coh.ta (‘rice - its taste is good’)
/pammasi( )cotha/ = pap (uy) mas i coh.ta (‘the rice taste is good’)
In the examples of this book a hyphen is used in the Romanization as an unobtrusive way to show
the first layer of internal structure of some of the words. In the citation of separate forms, the hyphen
sometimes shows the direction of attachment: -e must have a stem in front and mek- must have an
ending attached, as in the word mek.e ‘eats’.
1.9. External punctuation.
Koreans borrow English and Japanese punctuation freely. The standard practices accord, more or
less, with the current American usage. Parentheses, commas, and quotation marks are seen more
frequently than semicolons or colons. Sentences usually end with a period, but question marks and
exclamation points are also frequent, though their usage is not consistent. Korean questions are
typically marked by something specific in the sentence, such as the final postmodifier ya in ••• (ha)nun
A Reference Grammar of Korean
PART 1 21
ya, usually spelled as an unanalyzed ending (ha)nunya, and the interrogative ending -(su)p.nikka in
- hap.nikka. For questions so marked there is no need for a question mark, though many writers put
one in, anyway. When the interrogation is otherwise unmarked in speech, a rising intonation indicates
a question, and such questions are appropriately written with a question mark: Kim sensayng? Pap
mek.e? If the question mark is reserved to mark only those questions that have a rising intonation, we
can write the distinction that is heard with indeterminates, i.e. words that have both interrogative and
indefinite meanings: Nwu ka wass.ey yo. ‘Who is here?’
Nwu ka wass.ey yo? ‘Is someone here?’
Koreans now ordinarily write both these sentences with a question mark, just as at one time they would
have written them with a period (or no punctuation); the spoken distinction is lost in the writing.
In this book the question mark is used only for sentences with rising intonation. The period is
used at the end of statements and questions with falling intonation, but it is usually omitted when the
sentence is cited as an example with the English translation immediately following. See §2.8 for other
intonations that might be marked.
In some cases, the grammatical analysis used in this book is more detailed than that reflected in
the Hankul spelling systems. As a result, there are some word divisions that produce spellings at
variance with the prescriptions of Korean grammarians, such as -un ya for Hankul -unya, -urn ulo for
Hankul -umulo, -urn ey for Hankul -umay. The grammatical lexicon of Part II carries most of the
usual Hankul forms with cross references to the spellings I prefer: -unya -* -un ya.
In one or two other cases I have chosen to regularize forms that differ from those favored by the
Korean grammarians. I prefer the colloquial -e as the shape for the infinitive after stems ending in
-aC- (tat.e) rather than the literary/dialect version -a favored by the grammarians, though I realize
many younger speakers are tending to model their speech after the spelling. I prefer the colloquial hay
as the infinitive of ha- rather than the literary ha.ye favored by grammarians. But the literary forms
are discussed and cross-referenced. Another difference of opinion, more controversial: I do not
recognize the validity of a distinction between -te- and -tu- in certain retrospective endings, and
accordingly I treat all cases of -te- as literary or dialect variants of -tu-.
1.10. Alphabetization.
In the alphabetization employed in Part II, all superscript letters (*- ”••• s - h - y -) are ignored
except where entries are otherwise identically spelled, and the same is true for -q. The other letters
have their usual English order aceghiklmnopstuwy and the digraphs (kk, ng, wu, ey, ...)
are alphabetized by their component letters, as if the words were English. Vowel length is ignored
except for words that are otherwise spelled identically: the word with the short vowel comes first.
What about the alphabetization of words written in Hankul? There are several different orders in
widespread use, and the student may feel that each dictionary maker is plaguing him with new whims.
In general, the schemes fall into two types. The first is most widely used in South Korea, with three
variations on what to do with the geminates; the second type is official in North Korea.
(1) k (kk) n t (tt) I m p (pp) s (ss) -/ng c (cc) ch kh th ph h
a ay ya yay e ey ye yey o wa way oy wu we wey wi yu u uy i
(la) Ignore double consonants except where entries are otherwise the same.
(lb) Ignore initial double consonants except where entries are otherwise
the same, but keep a difference for final double consonants analogous
to that of the singlets:
k kk ks, n ns, 1 lk lm Ip is 1th lph lh, p ps, s ss
(lc) Recognize double consonants both initially and finally; make separate
places for the initial geminates (as in parentheses above), and keep
the final geminates in the order shown in (lb).
(2) n 11 m p s -ng c ch kh th ph h kk tt pp ss cc -
a ya e ye o yo wu yu u i ay yay ey yey oy wi uy wa we way wey
22 PART I
A Reference Grammar of Korean
In the latter system the circle symbol is treated in the consonant order (after s) only when final in a
syllable and pronounced -ng. When initial it is treated as a carrier for the vowels, and all the vowel-
initial words come at the end of the dictionary. In the South Korean system the symbol is usually
treated as the same, for alphabetizing, whether it is final or initial. (In the 15th century two slightly
different symbols were used and in certain instances the velar nasal ng could begin a syllable.) The
North Korean sai phyo (•••’), like our Romanized ™q, is ignored except when alphabetizing words that
are otherwise spelled identically. In the south the written sai sios is usually treated just the same as an
etymologically final s.
The names of the Hankul letters referring to vowels are simply the vowels themselves. The names
of the consonants are two-syllable nouns that echo the consonant at the beginning and the end: ki(y)ek,
niun, tikut = /tikus/, liul = /iul/ or /niul/, sios, iung, ciuc = /cius/, chiuch = /chius/, khiukh =
/khiuk/, thiuth = /thius/, phiuph = /phiup/, and hiuh = /hius/. These are the traditional names (Cf
C en Cayho 1961). In North Korea the traditional names for k, t, and s have been normalized to kiuk,
tiut, and sius, so as to fit better with the others. The geminate letters are named with ssang -
‘pair(ed), double’: ssang ki(y)ek / kiuk, ssang tikut/tiut, ssang sios/sius, ssang ciuc. There are no
special names for consonant clusters at the end of an orthographic syllable: -Ik is called liul kiyek
pat.chim. The traditional names for the first few letters come from the examples of the sounds in
initial and final position given by Choy Seycin in the introduction to his 1527 Hwunmong cahoy, a
dictionary of Chinese characters, and that accounts for the anomalous “ki-yek”, “ti-kut”, and “si-os”:
the latter two terms end with the non-Chinese words for ‘end’ (-* modern kkuth) and ‘clothes’, for
there were no Chinese syllables ending in -t (Middle Chinese -t Middle Korean -Iq). The symbol
representing the velar nasal was exemplified as i-ung, showing that it was not be pronounced at the
beginning of a syllable, and a different symbol (the circle without a tick at the top) was used for zero.
But that was written only at the beginning of a syllable in the readings of the dictionary, for the zero
final used by the earliest texts to end the vowel-final syllables was omitted, as had become customary.
The modern system uses the tick-topped circle, reshaped as a kind of teardrop, for both the zero initial
and the nasal-velar final. (The tick is often omitted and the teardrop flattened to a circle or an oval.)
Choy Seycin recognized that the letters kh, th, ph, c, ch, z [the triangle ^], zero [the circle], and h
(which he put in that order) were not heard as such at the end of a syllable and he cited examples only
in initial position, using the high front vowel: khi, thi, phi, ci, chi, [zji, i, hi. There was no syllable for
khi among the traditional Chinese readings, so he used the word for ‘winnow’ as a “non-Chinese”
word, though it may well be taken from a Chinese reading outside the mainstream of borrowings. The
terms khiukh, thiuth, phiuph, ciuc, chiuch, and hiuh seem to be modern neologisms, and North
Korea uses khu, thu, phu, cu, chu, and hu. The very earliest way to speak of each Hankul symbol
was probably to pronounce it, with the minimal vowel u supplied if needed. That is indicated by the
statements in 1451 Hwunmin cengum enhay of the type K nun ‘as for K’, in which the focus particle
appears in its postvocalic shape. Variant names for some of the letters turn up in 1881 Ridel (xviii):
miwom for mium, piwop for piup, eyas for ciuc, and ihwoyng for iung, which Ramstedt (1939:3)
gives as ihong = ihung; the version in 1874 Putsillo is the equivalent of ihoyng, while 1874 Pyankov
called the symbol ihyang. Ridel’s Romanization indicates that liul was to be pronounced with an initial
[r], but 1874 Pyankov gave the name of the letter as the equivalent of liur with a final Russian [r]
followed by the “hard mark”. 1883 Scott has the equivalent of tikkut for tikut and of ngiung for iung.
There are also names for the obsolete symbols of Middle Korean. The term iung is used to refer
to the zero initial (the circle O), and the symbol for the velar nasal Ois called yeysq iung ‘ancient
iung’. The triangle ^ that represents MK z is called pan sios ‘half s’ or samkak hyeng ‘triangle’; the
symbol for W tl (p with a little circle below) is called kapyewun piup ‘light p’ and the variant with m
instead of p (□) is called kapyewun mium ‘light m’; the symbol for q o , intended to represent a
glottal stop, is called yelin hiuh (= /hius/) ‘incomplete h’. A double zero oo, called ssangq iung was
used in writing a few forms with yGy and yGi from causative and passive verbs made with the
formative -Gi-, to make sure they were not taken as yy and y.i (= lyyil). Attested: "hoy 'Gi- ‘cause to
A Reference Grammar of Korean
PART I 23
do’ in the forms "hoy 'Gye (1451 Hwun-en 3) and "hoyGywo'm ol (1462 *Nung 6:87b); moy ’Gi- ‘get
tied’ in the forms moy'Gywon (1462 *Nung 1:43b, 55b), moy'Gi'no.n i ’’la (1447 Sek 13:9b),
moy'Gywu'm i (1447 Sek 6:29a), moy'yGwom 'kwa (1462 'Nung 6:28b) and moy'Gywo'm ol (1459
Wei 18:52a), moy'Gye (1459 Wei se:3b), elk-moy'Gi'ta (1447 Sek 13:9b); muy'Gi- ‘get hated’ in the
forms muy'Gin ( ? 1468 Mong 19b) and, according to LCT, muy'Gywon ([1467-*] 1517 Sapep 5b) -
but that is "muy'ywon in the version printed by Kwangmun-kak in 1979 (is this the 1543 edition?).
This was a clever extension of the device for writing MK G indirectly by not adding y- after -y or -i.
The other device was failing to link a preceding l or z with a following syllable y- or certain common
cases of if-). Geminate hh, called ssang hiuh oo, was used to write the voiced-h initial in the
prescribed readings of Chinese words and in representing the verb stem °hhye-, an intensive form of
°hye- ‘pull, drag’.
The letters that are joined to form a Korean syllable are sometimes found isolated, as when
ki(y)ek, niun, tikut, ... are used to mark items in a list, like English “A.”, “B.”, “C.”, .... As isolated
symbols the letters are called ttan ~ ‘separate •••’ or ‘free-standing •••’. In Middle Korean texts several
of the isolated letters are used for special purposes. The genitive particle s was often attached to the
preceding or following syllable, but it also appears as the sai sios A between two Hankul syllables or,
more often, after a Chinese character or its Hankul pronunciation. In addition to s itself, assimilated
variants were sometimes written as isolated letters: ttan pan sios ‘free-standing z’ ^ is found in
["HHWUW] z 'nal (1445 'Yong 26) ‘later days’, ttan kiyek ‘free-standing k' ~] shows up in ngwang k
"swon-toy (1447 Sek 24:6a) ‘to the king’, and ttan niun ‘free-standing n’ L_ appears in "wuy n "nim-
kum (1586 Sohak 6:40a) ‘the king of Wei’. In 1451 Hwun-en a free-standing W (ttan kapyewun piup)
tl was written for the genitive after Chinese words that were given prescribed pronunciations with a
hypothetical final consonant to represent the labial semivowel of the Middle Chinese versions. The
single bar that represents y in the digraphs of modern Korean was the glide of diphthongs ( ey, ay, oy,
uy) in Middle Korean, and it appears isolated as ttan i ] to write the nominative particle i after a
Chinese character or its prescribed pronunciation in Hankul when that ends in a vowel, causing the
MK particle to lose its syllabicity and shrink to the glide y. The same reduction took place for forms of
the copula i- : "THAY-"co ’yn ’ kwo't ol (1447 Sek 24:52a) ‘ ... that he was the prince’, ciN-ZYE ‘yn
cyencho ’y la (1465 Wen 1:1:1:47a) ‘ ... is the reason it is true’; nim-kum kwa [ssin-"hha] ’yl s oy
(1632 Twusi-cwung 6:32a) ‘it being a matter of king and court ... ’. But when the Chinese character
was pronounced with a final i or y the glide on the reduced forms was unpronounceable, so that the
copula stem was totally lost, leaving the modifier forms as ’n and ’l(q), written with ttan niun 1_ and
ttan liul £, as in ku cin-'ssilq s ti ’n 't ye (1464 Kumkang 87b) ‘is that the true wisdom?’ and hon
"THYEY 7 's ila (1482 Nam 1:39a) ‘they are a single body’. Yet in the same situation, despite the
unpronounceability of •••ly or - Yy, the nominative particle was often written with ttan i anyway: "Lly
'ngwok 'kwa "twolkhwa [= "twolh kwa] oy tal'Gwo.m i "ep.swu'toy (1475 Nay se:3a) ‘reason does
not have the differences of jade and stone’; ku ~ccwoY y 'stwo tye ’y sye ne'mu.l i ’ la (1463 Pep
4:83a) such sin is greater than that; zye loy y 'cukcay ssin-'luk "nay'sya (1463 Pep 6:97a) ‘the
tathagata at once displayed his supernatural power and ... ’.
2.0. Sounds.
The sounds of Korean are described in terms of phonemes and components (§2.1), followed by a
more detailed discussion of the phonetic articulations and the acoustic impressions they impart
(§2.2-4). Syllable structure and intersyllabic strings (§2.5-6) are restricted by various assimilations and
incompatibilities that have arisen through the years. A variety of sequence variants (§2.7) are found, as
well as conflicting judgments on the standardization of competing variants (§2.8). The principal
intonation patterns are summarized in §2.9.
2.1. Phonemes and components.
For the variety of Korean speech that we take as our point of departure, the phonemes and their
components are as shown below.
24 PART I
A Reference Grammar of Korean
Simple vowel nuclei
FRONT BACK
ROUNDED UNROUNDED
ROUNDED
UNROUNDED
HIGH
-
i
U
wu
MID
oy
ey
e
0
LOW
—
ay
a
—
Complex nu
clei
wi
-
-
-
- yu
wey
yey
we
ye
- yo
way
yay
wa
ya
“ ™
Consonants
Lax
Reinforced Aspirate
Nasal
Liquid
Labial stop
P
PP
ph
m
Dental stop
t
tt
th
n
1
Alveolar affricate
c
cc
ch
Velar stop
k
kk
kh
ng
Spirant
s
ss
h
2.2. Vowel descriptions.
The vowel chart displays a nearly maximal system. In standard Seoul speech oy (mid front
rounded vowel) is not distinguished from the diphthong wey, and there is no need for the front
rounded category at all. The distinction between (pyeng i) toycinta (= tocinta) ‘(the illness) worsens’
and (ce salam i) tweycinta (< twie cinta) ‘(he) drops dead’ is orthographic for Seoul speakers; many
pronounce the first syllable of twayci ’ta ‘it’s a pig’ the same way, as /twey/. Moreover, in rapid
speech the w will often drop, leaving /tey/ as the first syllable of all three phrases. There are very few
words with wey or way, for that matter, and they are often confused in spelling with the many words
that contain oy. Examples are kwey ‘box, case’, kweyto ‘railroad track’, kkweyk ‘with a shout’ (=
kkwayk), kkweynta ‘strings, puts through’, weyn ‘why, what’; way ‘why’, waykhong ‘peanuts’,
kkway ‘extremely’, insway ‘printing’, yukhway hata ‘is delightful’, and a few others.
Some linguists would move wi from the group of complex nuclei into the high rounded category
of simple vowels, since many speakers tend to pronounce wi as a long monophthong [ii] rather than the
more common diphthong [ui], coming from an earlier wuy [u|]. For most speakers, the phoneme /w/ is
represented by simple lip rounding, with the tongue position largely determined by the following
vowel: wi [ui], wey [oe], way [3e], we [oe], wa [oa]. The phoneme lyl is usually high [i] regardless of
the following vowel. Some speakers of the Kyengki area have a full range of rounded front vowels,
monophthongizing wi [ii], oy (and probably wey?) [6:], and way [5:] alike, at least when these are
historically long.
The vowel e has two markedly different allophones in Seoul speech: higher [e] when long, and
lower [e] (= IPA inverted-v) when short. Many southern speakers give the vowel a slight internal
rounding, as if scooping the back of the tongue with a spoon, when it is at the end of a syllable and
especially before a pause. In Seoul speech there is a strong trend toward a rounded but unprotruded
[oj. Other vowels are less noticeably different when long or short, but in general the long high and
MID vowels tend to be higher than the short ones, and the long LOW vowels (if anything) somewhat
lower than the short ones.
A Reference Grammar of Korean
PART I 25
In the noninitial syllables of Seoul colloquial speech, wu is widely substituted for short o after a
consonant, and (much less often) the syllable yo gets pronounced as yu, though I believe that is not
true of consonant + yo. In very common words, especially endings and particles such as -ko and to,
the Seoul colloquial forms (-kwu and twu) can be regarded as standard relaxed speech. But when wu is
used for o in initial syllables it is considered substandard; twun for ton ‘money’ sounds vulgar. There
are situations where that sort of speech can be effective: Nwungtam ici! ‘It’s a joke = I’m joking!’ (<
'nongtam). The substitution of wu for [w]o can take place for each word independently: Coh.ko
malkwu yo, Coh.kwu malkwu yo, and even Coh.ko malkwu yu and Coh.kwu malkwu yu, as well
as Coh.kwu malko yo and (?) Coh.ko malko yu.
A similar raising of the back mid vowel e to u and the front mid vowel ey to i is also heard in
noninitial syllables, and less commonly even in initial position. You may hear ups.ta (and, for some
reason, even wups.ta) instead of eps.ta, cukta for cekta, -usi yo for -usey yo, etc. There are speakers
who raise a to e in variant versions of certain words: - henthey for ••• hanthey ‘to’, mulle for molla
‘dunno’, etc. (Cf the remarks on -tun/-ten in Part II.) Some raised vowels are considered standard.
But eti ‘where’ probably comes directly from MK e'tuy rather than from e[nu] tey ‘what place’.
Throughout much of southern Korea the vowel ay is distinguished from ey poorly, if at all. In
Seoul speech the distinction is seldom maintained in noninitial position, though speakers are aware of
it. The two expressions sey kay ‘three things’ and seykyey ‘world’ are usually pronounced identically
as /seykey/. One young Seoul linguist says of this pair of words “I can make the distinction, but I
don’t”. (The name of a Honolulu restaurant /phainpeylli/ sounds like “Pine Belly”, but it turns out to
be “Pine Valley”!) Some of the homonymy that results makes it harder to identify morphemes: the
string /ettekhey/ may represent either etteh.key or etteh.k’ ’ay (contraction of etteh.key hay); and
/epsseyyo/ may be intended either as eps.e(y) yo ‘lacks it’ or epsay yo ‘does away with it’. With
further raising of /ey/ to /i/ the expressions elin i ‘young one’ and elin ay ‘young child’ converge. In
the areas where ay and ey have merged, confusion is avoided in various ways. To keep nay kes ‘mine’
distinct from ney kes ‘yours’, for example, Taykwu speakers use na kes and ne kes. Other word pairs
that might be expected to cause difficulty, such as kay ‘dog’ and key ‘crab’ are kept distinct by raising
the mid vowel; in Taykwu ‘dog’ is key and ‘crab’ is kl. And among younger Seoul speakers one hears
“ney kes - ni kes” = nay kes - ney kes ‘mine - yours’; raised vowels are also heard in ni ka =
ney ka ‘you’ and even ci ka = cey ka ‘I [formal]’. Back formations (hyperurbanisms) are sometimes
heard from southerners who merge ey and ay, as when cay-il is said for cey-il ‘number one’; one
speaker (recorded on KBC 27:6) said selmayng = selmeyng < selmyeng ‘explanation’.
The dialect of Kimhay, on the southern coast, has a minimal system of vowel distinctions. There
are six vowel phonemes:
FRONT (UNROUNDED) BACK UNROUNDED BACK ROUNDED
high i u/e wu
low ey a o
The quality of the u/e vowel varies from high to high-mid (depending on what is around it), as does
the mid to low-mid quality of ey and o. The Kimhay vowel ey is cognate with standard (written)
Korean ey, ay, oy, way, and wey; the vowel i is cognate with both i and wi in the standard written
language.
The dialect of Ceycwu island (LSN 1978) retains certain features of earlier Korean that were lost
in other dialects, notably the vowel traditionally called alay a (= a lay 6), which the Ceycwu speakers
pronounce as a lower back rounded vowel (“open o”), though the Middle Korean equivalent functioned
as a low back unrounded vowel with the closed o (= wo) as its rounded counterpart. As in other
dialects the earlier diphthongs represented by our spellings ey and ay are monophthongized as front
unrounded vowels, uy has merged with i, and the usual pronunciation of wuy is i after a consonant
26 PART I
A Reference Grammar of Korean
but after a vowel or pause it is a front rounded vowel, like some versions of standard wi (which comes
from earlier wuy), though the more common Seoul pronunciation is a front rounded glide followed by
an unrounded high vowel, rather like the sound in French “huis” or “lui” rather than that of French
“oui” (English “we”) or “Louis”. As in many southern dialects, the distinction of historic ay (and oy)
from ey is maintained poorly, if at all, though transcriptions may lead you to think otherwise. The
unrounded vowels u and e are kept apart, and e is pronounced as a shwa [a], rather than the currently
popular Seoul pronunciation as a low back rounded vowel [a], the value of which is preempted in
Ceycwu by the old vowel o which in other dialects has merged with a in initial syllables and u in other
syllables, with further assimilations in particular phonetic environments. Ceycwu retains a syllable yo
which was merged with ye or ya in the central dialect of the early Hankul texts, though provision was
made for the syllable (by a “double low dot”) in the scheme of the letters
The old diphthongs now monophthongized in most dialects are still heard as diphthongs in various
parts of Hwanghay (Kim Yengpay 1984:297), where the pronunciation faithfully follows our spellings
of say ‘bird’, kay ‘dog’, key ‘crab’, oy ‘cucumber’, and oy- ‘memorize’! The 1898 Tayshin dictionary
has good examples of ai and ay for the diphthongs, including nwukwuy for ‘who’ (■*- nwukwu i) and
sorai for ‘sound (etc.)’ = NK solay(ki) < MK swo'loy (> Seoul soli).
Most speakers devoice (= whisper) high-vowel nuclei (i wi u uy yu wu) in an environment with h
or s (or consonant + h or s) on one flank and a simple obstruent (p t c k s h) on the other: puchica,
khuta, chwupta, sik.kwu, huksayk. kupsa, swita, swipta, swipsita, hu^ta. Some speakers devoice
the mid and low vowels in the same environment: haksayng, saykssi, thokki, thepthephata. But if
there is only one vowel in the phrase it is usually voiced: sip, huk, swuch; hak, thek.
2.3. The pseudo-vowel uy.
The spelling uy is preserved from an earlier stage of the language, when ey and ay were
diphthongs and wi was pronounced wuy [ui]. At that time uy was pronounced [iij, but with the
monophthongization of the other diphthongs uy, too, was monophthongized to /u(:)/ or /i(:)/,
depending on the dialect. In Seoul the traditional pronunciation of the monophthong is u at the
beginning of a word and i in noninitial position. But when it is initial in a word, some people attempt
a reading pronunciation of the diphthong and say ui (in two syllables); that pronunciation has been
rapidly spreading among younger speakers, partly as a result of schooling. In the traditional Seoul
versions, the word uykyen ‘opinion’ is pronounced ukyen and the word cwuuy ‘attention’ (which has
for its last syllable the same Chinese morpheme ‘one’s will’ that is the first syllable of uykyen) is
pronounced cwOi. In an obvious compound noun like Sin-Uycwu, name of the city of “New Uycwu”,
two versions are heard: sinicwu, treating the sequence as one word, and sinucwu, treating it as two
words (but with no overt juncture separating the two). But mu-uymi ‘meaningless’ is usually
pronounced muimi though uymi ‘meaning’ is Omi; mu- ‘lacking’ is bound and functions as a prefix,
whereas sin(-) ‘new’ is quasi-free and can function as an adnoun. (There are no reports of the
pronunciation *mu-umi > [mu:mi] and it has probably never existed.) The odd-looking word uyuy
‘meaning’ is pronounced ui.
The spelling uy, like the spelling oy, is left over from a period when Korean had phonetic
diphthongs. The diphthongs ay and ey began to get monophthongized as front vowels from the late
1700s, but (w)oy, wuy, and uy seem to have persisted into the 19th century, at least. Present-day
Korean has reduced the back rounded vowel of the older wuy to the semivowel w and lengthened the
palatal glide y into the high front vowel i to yield wi, but some speakers use a front rounded
monophthong [u] at least in certain environments. A similar front rounded monophthong [6] is used for
older woy > oy by some speakers (in some environments) but the general pronunciation reduces wo to
w and fronts the vowel to ey, so that the syllable merges with the articulation of wey, which was
separate from (w)oy earlier. And there are speakers who, at least sometimes, treat (w)oy as way, in
Seoul as well as (Kim Yengpay 1984:67) in Phyengan. In Phyengan wuy developed into wi but at the
end of a polysyllable it often dropped the ~y to become wu (ibid): Tang-nakwu = Tang-nakwi
‘donkey’ (< la'kwuy), ppey-takwu = ppye-takwi ‘bone’; pakhwu = pakhwi ‘wheel’ (< pakhw^by
A Reference Grammar of Korean
PART I 27
< pa'hwoy), pawu = pawi ‘rock’ (< pawuy < pa'hwoy). There is no monophthong articulation that
could replace the diphthong uy without merging it with either u or i, and many dialects have followed
the expected course of pronouncing it as i. But in Seoul the word-initial uy has resisted that
development and monophthongized to u, instead. It is possible that in the word-initial position the
diphthong version uy has simply persisted for some Seoul speakers, as certain linguists would prefer,
but I believe it is a back formation based on the spelling, as I have said. There are bits of evidence for
the traditional Seoul treatment as early as the end of the nineteenth century, when Tayshin in his 1898
Korean-Russian dictionary wrote the equivalent of /usimi/ for uysim ‘doubt’ (with the accretion of the
suffix -i that is widely attested in North Korea), though he wrote /uisik/ for uysik ‘clothes’. The
Phyengan development for uy is i but the particle is either ey (as in Seoul) or u (as in Cincwu, Mkk
1960:3:31). An early doublet: i'ki- (1445 *Yong, 1449 Kok) = i'kuy- (1447 Sek etc.) ‘win’.
2.4. Consonant descriptions.
The lax obstruents are weakly articulated; in initial position they are released with a slight puff of
local breath (in contrast with the heavy breathing of the aspirates) but are less tense than their English
counterparts: [b‘] for p-. Between typically voiced sounds (vowels or semivowels, nasals, the liquid),
the lax consonants are lightly voiced in rapid speech, as [a:nda] for anta ‘knows’. Lax consonants do
not occur after other lax consonants; our spellings pk, ks, etc., automatically represent the phoneme
strings /pkk/, /kss/, etc. (see §2.6). The lax consonants p t k occur in syllable-final position and there
they are unvoiced and unreleased [b 1 d 1 g 1 ]. 1881 Ridel was quite hesitant about the voiced allophones
of the lax consonants: “sometimes - but not ordinarily” voiced.
The aspirated consonants are begun with a lax articulation (contrary to some descriptions),
frequently velarized, and followed with heavy aspiration that is often accompanied by velar friction:
[b h a] or [b x a]. The aspirated obstruents are never voiced. The phoneme /h/ is frequently “voiced”
(murmured) or dropped between voiced sounds. It is more commonly glottal than velar, but a velar
variant occurs, as well as a velarized glottal. Initial hw, and especially hwu, is often pronounced as a
bilabial fricative [F] by many speakers. Ridel (1881:xv) transcribed the aspirates as “hk hp ht tch”, and
Underwood (1914) used “hk hp ht” but “ch” (writing the lax c as “j”). This follows the traditional
Romanization of Burmese and has certain virtues in handling morphophonemic problems of Korean
(see Martin 1982); Starchevskiy (1890) also wrote the aspiration before the consonant (King 1991a).
Roth (1936:6) refers to a system like Ridel’s as the “French transcription” (with “tj” for the simple Id
and the geminates “kk pp ss ttj”) and he refers to a somewhat similar system except for “htsch” (with
“tsch” for the simple Id and the geminates “gg dd bb ss dsch”) as the “German transcription”.
The reinforced consonants are pronounced with great muscular tension, both locally and through
the entire vocal tract. The laryngeal tension continues on into the vowel, which can be described as
“laryngealized” (somewhat gargled), and the effect of the release is a clearcut popping similar to that
of glottalized consonants but with no separately heard glottal release, so that they sound unlike the
glottalized consonants of certain languages native to North America. The tense unaspirated stops of
French are somewhat similar to these sounds. The reinforced consonants are never voiced.
The aspirated and reinforced consonants occur only as syllable onsets, but the syllable itself need
not be word-initial. When they are internal in the phrase there is no distinction between strings like
appha and apha, or apppa and appa. In rapid speech only the latter occurs, and we take that to be the
phonemic norm; in slow speech the stop is anticipated with closure of the preceding open syllable.
These two tempo-controlled articulations, which I call compressed and conflated, are variants
regardless of the constituency of the phrases in which they occur. The morpheme boundaries may be
a-pha, aph-a, or ap-pha (in basic form also ap-ha or ah-pa), but the compression and conflation will
take place automatically for each, as the tempo changes, giving two phonetic realizations for each
phoneme string /apha/ and /appa/, regardless of their meaning, if any. (See Martin 1982 and 1986,
where mention is made of similar variation in English words like “upper” and “cookie”; also §2.10.6.)
The nasals are fully voiced, and at the end of a syllable somewhat long. Initial m, especially
before (w)u and o, often has an oral release [m b ] or even [ m b]. Initial n, especially before i or e (as in
28 PART I
A Reference Grammar of Korean
ney ‘yes’), sometimes has a similar oral release [n d ] or [ n d]. (Cf Lukoff 1954:4, 11; Martin ? 1991:
nl4.) Ramstedt (1939:14) cites NK dialect forms “ n dui n lui; duin luin” for ‘four’ (neys), perhaps a
variant of the [doy] reported by 1900 Matveev (King 1988b:309) and [noi] in 1898 Tayshin; compare
the Hamkyeng form cited by Kim Thaykyun as nei. The velar nasal ng does not occur at the beginning
of a word, or after pause, but it can occur as the onset of a noninitial syllable as a result of the
consonant “liaison” described in §2.5. In the strings -ngi and -ngy- the velar nasal is fronted and
weakened, often disappearing into a nasal y [I], or vanishing completely as in the northern dialect of
1902 Azbuka (and sporadically even in Seoul); King (1988:301-2) reports the phenomenon in 1898
Tayshin, with the nasal vowel remaining, and 1900 Matveev, with no nasality written.
The liquid /l/ occurs after pause only in recent loanwords, and at the beginning of a word only in
recent loanwords or in words that do not occur at the beginning of a phrase, such as particles.
Internally, HI occurs both as a syllable onset and as a coda. When the liquid is syllable-initial, and not
preceded by another 1, the tongue tip is quickly flapped against the very front of the alveolar ridge, so
that the articulation sounds like the single-tap Spanish r or (rather) the somewhat more liquid Japanese
r. At the end of a syllable, the 1 is unreleased and the tongue curls up around and beyond the point
where it would have been released, producing a sound that English speakers hear as [1], The string /ll/,
syllable-final 1 + syllable-initial 1, sounds like a long 1 to a speaker of English, but without the
velarization that colors the “dark 1” of many English speakers. But Lukoff 1954:40-2 seems to feel that
/ll/ is phonetically short like Ini rather than long like Inn/ [n:n], and some native linguists have
expressed a similar feeling. Perhaps that perception (or misperception) is due to the existence of an
initial [n~] but no initial [1-], for /!•••/ is realized as the flap allophone [rj. Yet at the same time Lukoff
(1954:9-10) describes the articulation of /!/ after /!/ as a “pre-flapped [1]” which he writes as [ d l].
Often the Korean 1 (and even more often the 11) is somewhat palatalized, especially before i or y; that
is, the center of the tongue is humped at the same time that the tip of the tongue is making the primary
articulation. There have been various reports that the flap allophone is used by northern dialects in
place of the lateral used in the south before pause or a consonant, e.g. Ramstedt (1939:11) says the
allophone of the liquid before a consonant is [1] in South Korea but [r] in North Korea. And there is
evidence for that in Russian transcriptions from the 1800s and early 1900s; the Cyrillic spellings of
1902 Azbuka have [r] for the liquid in -1C- clusters. King 1987 says that the North Hamkyeng dialect
preserved in the USSR has [r] word-final and before obstruents. He tells me that such speakers
pronounce -11- as a short lateral, in contrast with [r] for the simple -1-. In the expression il.il
liphothe ‘daily / everyday reporter’ you can hear a sequence of lateral + flap [iril( )ri—] but that
probably represents a juncture; with totally suppressed juncture the pronunciation is [irilli—].
In rapid speech, owing to the consonant “liaison”, the strings mh, nh, ngh, and lh often occur as
syllable-initial clusters, instead of as a syllable-final consonant + the onset h-. Here the h is “voiced”
(murmured) and pronounced as a breathy release of the somewhat shortened nasal or liquid, which has
its “r”-like sound, as noted by Ridel (1881:xiv). More common is a variant version which drops the h
completely; see §2.7.3.
The distinction between the two sibilants s and ss is often difficult to hear, and many Koreans,
especially in the south, appear to lack the distinction in their local speech. Though minimal pairs can
be found, the functional load of the distinction seems to be fairly low, especially in noninitial position.
To an American ear, the best description might be this: s is something LESS than what you expect of an
“s”, and ss is something MORE. There is a fuzzy “lisp”-like quality to the lax s. If you hear a clearcut
“s”, it is probably the reinforced ss. If you hear an “s” that you can’t make up your mind about, an
“s” that seems to have something missing, it is probably the lax s. In the Seoul speech of some, the lax
s before i and wi is regularly palatalized [z‘ y i] or even palatal [?‘i], and that helps distinguish si from
ssi. But in the speech of others, the ss is also palatalized before i and wi, so that palatalization is not so
reliable a guide as the lag, after the plain s, in voicing the vowel, a lag that is indicated in our phonetic
transcription by the inverted apostrophe used to symbolize lightly aspirated release.
The affricates are palatalized throughout the south, and the stop that begins the affricate is
sometimes a palatal stop, which identifies the phoneme more perceptibly than the sibilant release,
A Reference Grammar of Korean
PART I 29
especially the reinforced cc, which tends to suppress the sibilant along with any hint of aspiration. For
many Seoul speakers the palatalization is weak or absent before back unrounded vowels, and in the
north the nonpalatalized affricate is frequently heard before all the vowels. Figulla 1935:103 says Id
before /a e o wu/ is the dental affricate [ts], elsewhere (i.e. before /i ey ay wi wey way oy/) the palatal
affricate. After a vowel and before a back vowel, the voiced version of the nonpalatalized affricate is
sometimes weakened to just [z]. Good examples of that can be heard from a female speaker on the
tapes accompanying KBC (Pak Pongnam 1969): ize = ic.e ‘forgot’, ezey = ecey ‘yesterday’,
kuzekkey = kucekkey ‘day before yesterday’, sazen = sacen ‘dictionary’, paykhwazem = payk.hwa-
cem ‘department store’; ui(d)za = uyca ‘chair’, mo(d)za = moca ‘hat’. Despite the description
implied in several treatises, it is quite rare to hear [z] in any modern dialect as an allophone of Isl;
when [z] or [ d z] is heard, it almost always represents Id. But there is evidence for a [z] allophone of
Isl in earlier Hamkyeng speech, as heard by Putsillo, Matveev, and Tayshin. And Lukoff (1954:8) says
of Is/ “After /m, n, ng/ and before a vowel, it may have a weak and sometimes slightly voiced variant
[z].” On [z] see also §2.11.4. Ridel, who only hesitantly admitted voiced versions of p t k, flatly states
(1881 :xvii) that there was no [z],
2.5. Syllable structure and consonant liaison.
The Korean syllable is a phonetic entity that is automatically predictable in terms of a string of
phonemes: it consists of an initial (the onset), a medial (the vowel or diphthong), and a final (the
coda). The onset can be zero or p pp ph t tt th k kk kh c cc ch s ss h m n -ng (-)!; -mh, -nh, -ngh,
-lh. The final can be zero or p t k m n ng 1. The Korean writing system, and our transcription, has
other finals in its “orthographic” syllables, but those are basic forms subject to reduction in the spoken
syllables. Modern Hankul uses the same symbol for initial “zero” and final -ng, though distinctive
symbols were once used; the final zero is left unmarked, though it was marked in some of the earliest
Hankul texts.
Not all possible combinations of initial, medial, and final occur. Medials beginning with y (yey
yay ye ya yu yo) do not occur after s ss c ch cc t th tt except as contractions from i + vowel (iey iay
ie ia iwu io) — see §3.5.(9), or in a few recent loanwords such as syassu (also syaccu and syechu)
‘shirt’ for which the less Japanese-sounding sassu is also heard. The vowel u does not occur after a
labial (p ph pp m) and that is why we can abbreviate the vowel wu to u after a labial, so that our
“pu” = p + wu, “phu” = ph + wu, “ppu” = pp + wu. In a similar way our “yu” = y + wu
since there is no y + u. Medials that begin with w (wi wey way we wa) do not occur after a labial
except as abbreviations of wu or o + vowel (wui wuey wuay wue wua oa), as in mwe(s) < mue(s)
‘what’.
Whenever possible, a syllable shuns the “zero” initial. Korean syllables like to begin with a
consonant. When a syllable with the zero initial is appended to a syllable with a final consonant, that
consonant shifts over to become the onset of the second syllable: pap + i is pronounced pa-pi, pang
+ ey is pronounced pa-ngey, pal + ul is pronounced pa-lul and sounds just the same as pa + lul.
When the second syllable begins with h, the result is what we would expect from our transcription: -p
+ h- -* -ph- etc. In rapid speech we even find -m (-n -ng -1) + h- -* -mh- (etc.), though
the h frequently drops in this position. When the first syllable ends in p, t, or k and the second syllable
begins with the same consonant, the first syllable loses its final and the initial of the second syllable
doubles: — p 4- p- -* — pp-, etc. (For the fact that — p + k— -* — pkk-, —t + k— -* -kk— , — t +
s- -* —ss—, etc., see the following section.)
How many distinct syllables does Korean have? It is difficult to answer this question precisely.
Suppose we figure that the initials (including zero and I-) are 20 (+ 5 in spoken syllables like -ngi,
-mha, etc. = 25), that the medials are 20 (omitting uy because of its limited distribution and doubtful
status as a phoneme), and that the finals (including zero) are 8. Then we have a minimum (20x20x8 = )
3,200 and a maximum of (25x20x8 = ) 4,000. From these figures we would perhaps want to subtract
the syllables that do not occur in spoken words, but many such syllables will in fact turn up in Hankul
30 PART I
A Reference Grammar of Korean
spellings, for one reason or another, and in addition Hankul has a number of orthographic syllables
with additional basic finals (such as ph th kk ps Is etc.) and some unusual syllables like sya eye
chyess, etc. There are undoubtedly some accidental gaps that do not occur in normal words. Perhaps
we are safe in saying that Korean has between three and four thousand different syllables, some of
which are fairly rare, and many of which are limited to certain sections of the vocabulary, such as
Chinese loanwords, inflected forms, and mimetic words (onomatopes).
2.6. Cluster restrictions.
When two syllables occur in uninterrupted sequence, fewer strings of consonants are found than
we expect. The occurring strings are shown in the following table. The line across the top shows the
end of the prior syllable; the column at the left shows the beginning of the following syllable. At a
point of intersection, an expected string is shown in lowercase boldface. An automatic replacement of a
morphophonemically expected sequence is shown in boldface italic. The notion of “expected” is with
respect to the analysis underlying the Romanization: when a syllable ending in -p is attached to one
beginning with -p, the string that results is syllable-initial pp-, pronounced as the onset of the second
syllable but optionally picking up a parasitic final -p (not the original -p) when the articulatory process
is slowed down. Syllable boundaries are assumed to be automatically determined at a given tempo. The
zero onset and coda are noted with the symbol “0”.
-P
-t
-k
-m
-n
-ng
-1
-0
p-
PP
PP
kpp
mp
mp
ngp
•P
P
ph-
ph
ph
kph
mph
mph
ngph
lph
ph
pp-
PP
PP
kpp
mpp
mpp
ngPP
•PP
PP
t-
ptt
tt
ktt
mt
nt
ngt
It
t
th-
pth
th
kth
mth
nth
ngth
1th
th
tt-
ptt
tt
ktt
mtt
ntt
ngtt
ltt
tt
s-
pss
ss
kss
ms
ns
ngs
Is
s
ss-
pss
ss
kss
mss
nss
ngss
lss
ss
c-
pcc
cc
kcc
me
nc
ngc
lc
c
ch-
pch
ch
kch
inch
nch
ngch
Ich
ch
cc-
pcc
cc
kcc
mcc
ncc
ngee
lcc
cc
k-
pkk
kk
kk
mk
ngk
ngk
lk
k
kh-
pkh
kh
kh
mkh
ngkh
ngkh
lkh
kh
kk-
pkk
kk
kk
mkk
ngkk
ngkk
lkk
kk
h-
ph
th
kh
mh
nh
ngh
Ih
h
m-
mm
nm
ngm
mm
mm
ngm
lm
m
n-
mn
nn
ngn
mn
nn
ngn
ll
n
1-
0-
mn
(nn)
ngn
mn
ll
ngn
11
1
This table can be regarded as a kind of filter, through which the underlying morphophonemic strings
that we expect to occur across morpheme boundaries (or orthographic pseudo-boundaries) are
converted to the phonemic strings that serve as input to the rules that tell the articulatory organs how
to realize the utterances. A native speaker of Korean unconsciously utilizes a filter of this sort, not
directly accessible to observation. The filter can, however, be generated by a set of rules that more or
less recapitulates the history of changing phonetic habits through the centuries, and it is possible that
the speaker creates his filter, or bypasses it, by making use of such a set of rules. Below are the rules
that account for the table, with a few notes on the historical developments that brought them into
existence. (The rules are a revision of those in the introduction to KEd. The notes are largely based on
Martin 1989 and the works cited there.)
Rules to convert morphophonemic/orthographic strings to phonemic strings:
A Reference Grammar of Korean
PART I 31
(1) Nasal lateralization. Change n to I when it is contiguous to 1.
(2) Liquid nasalization. Change 1 to n when it is after a consonant other than 1 or is after juncture.
(3) Nasal assimilation. Convert the oral stops p t k to the corresponding nasals mnng when a
nasal (m n) follows.
(4) Cluster reinforcement. Reinforce simple p t k c s to pp tt kk cc ss after an obstruent (p t k).
(5) Assibilation. Pronounce t as s before s or ss.
(6) Assimilation of apicals. Make the point of articulation of t and n be the same as that of a
following labial (p pp ph m) or velar (k kk kh): t -* p or k, n -* m or ng.
(7) Cluster reduction. Unless deliberately slowing the articulations, reduce three like consonants
to two (ttt -* tt); before h reduce two like consonants to one (tth -* th).
It will be observed that strings of lax obstruents are not permitted: the expected kp is
automatically replaced by kpp, etc. Since most cases of lax obstruent + reinforced obstruent (kpp, .. )
are the result of juxtaposing two basic lax obstruents (-k + p-, ... ), the default Hankul spelling is -k-
p- (... ) for all cases of /kpp/ (... ) except when the second element is clearly a form that has a basic
shape with initial pp- (...). The string /yakpalle/ can represent both yak palle ‘applies medicine’ and
yak-ppalle ‘is shrewd and quick’. We are reminded of the appropriate spelling in each case by the
recurrence of the forms in such unambiguous contexts as yak ul palle /yakul( )palle/ ‘applies
medicine’ and yak.ko ppalle /yakko( jppalle/ ‘is shrewd and is quick’. As in other cases where
morphophonemic decisions are called for, Koreans sometimes get confused and misspell words, either
in the morphophonemically safer direction (-kp-, ... ) or in their phonemic form (-kpp-, ... ) as they are
heard. (And the decision on where to divide a word into morphemes is sometimes in conflict with the
history of the forms; see §2.10.6 for examples.)
An interesting example is provided by the convergence of the two sentences pap to iss.e ‘there is
also (cooked) rice’ and pap tto iss.e ‘there is more rice’, which sound the same in rapid speech when
no pause is inserted: /papttoisse/. With pauses (“open juncture”) to distinguish the two sentences, they
sound different: pap to iss.e /paptto lisse/, pap tto iss.e /paplttoisse/. In slow and overdeliberate
pronunciations you may hear -tkk- and -tpp- instead of the usual reductions of -kk- and -pp- from
-tk(k)- and -tp(p)-. Some speakers feel that they also articulate -tcc- for the -cc- from -t c(c)- but that
is questionable. The pronunciation of -tss- from -t s(s)- is highly artificial; a genuine llssl is used by
many speakers in pronounced the loanword syaccu ‘shirt’ (also syassu) as [Sattsi], following the
articulatory habits of Japanese, from which the word was borrowed.
As earlier observed (§1.5), Chinese loanwords regularly reinforce the sequences — Is—, -lc— , and
-It- to —lqs—, -Iqc— , and - iqt— • For /lc/ and /It/ it is fairly easy to find non-Chinese words
without the -q-, such as the forms made by attaching the endings -ci and -ta to 1-extending verb stems
(§8.3.2), and compounds such as thel cangkap ‘wool gloves’. Words made up of Chinese morphemes
may appear in compounds without the reinforcement: sil-cakca ‘a reliable person’, chelmul-cen
‘hardware store’. For /Is/ it is not so easy to find contrasts, and I suspect there may be Koreans who
reinforce all cases to -lqs-. (To be sure, many Koreans simply fail to distinguish /ss/ from Isl in any
environment; we do not speak of them.) Yet certain types of compounds and phrases turn up cases
where I have observed /Is/ from at least some speakers. One textbook (M 1:1:70) indicates kyosilq se
for ‘in the classroom’ but that is pronounced kyosil se by speakers I have heard (such as J Yi Tongcay).
Examples:
pyel soli ‘unexpected remark’, pyel swu ‘extraordinary good fortune’, pyel salam ‘an
eccentric’; Sewul se ‘from Seoul’, Sewul si ‘the city of Seoul’, thukpyel si ‘special city [of
Seoul]’; pal-soy ‘informing on others’, naphal-swu ‘bugler’, sol-song namu ‘hemlock spruce’,
sal-son ulo ‘with bare hands’; mal sol ‘horse brush’, cil soth ‘earthenware kettle’, mal somssi
‘eloquence’, mal silqswu ‘tongueslip’, congtal say ‘skylark’, chel say ‘seasonal (= migratory)
bird’, thel sll ‘wool yarn’, cil sas-pan ‘small reed tray attached to an A-frame carrier’,
Hankwuk mal sensayng ‘teacher of Korean’; silsil-i ‘thread by thread, every thread’; hoth-pel
salam ‘shallow-minded person’; wul seyta ‘(clan is strong =) has a large family’, kkol
32 PART I
A Reference Grammar of Korean
sanapta ‘has an ugly face’, pul salunta ‘commits to the flames’, al sunta (+- su-1-) ‘emits
spawn, lays eggs’, nal sunta (*- su- < se-) ‘gets sharp(-edged), takes an edge’, nal seywunta
‘sharpens, puts an edge on’; twul sai ‘between the two’, but twulq cwung ‘of the two’ - 'Yi
Mayngseng tells me that he uses twulq sai more often than twul sai and that he thinks twul
cwung occurs but is rare.
In certain compounds made up of Chinese elements the reinforcement seems to be optional: x ipal(q)-so
‘barber shop’, 'ipal(q)-sa ‘barber’.
The assimilations and reductions described here take place across words and phrases, when the
phonetic cues to their boundaries are omitted, as often happens in normal rapid speech: Acik moluni?
‘You still don’t know?’ (/••• ngm—/), Wang sepang to kkamcak nollan mas! ‘The taste that surprised
Mr Wang!’ (ad for 3-minute “instant meal”) (/-ngn•••/), kot na onta ‘will come right out’ ( l—nn—l ),
ton ul naynta ‘pays the money’ /tonullaynta/, khunq 11 nass.ta ‘that’s terrible’ /khunnlllatta/.
2.7. Sequence variants.
There are certain types of variants which are widely systematic: a given sequence of phonemes for
which we always (or always within a morph or a word), find a variant of consistent shape. There are
also some which are less predictable, but also widespread, of a similar sort. Nine types of these
SEQUENCE variants are described below. See also the remarks on oy (§2.2) and uy (§2.3).
2.7.1. Precision variants.
In speech at a normal rapid tempo, n is replaced by m before p or m, and by ng before k:
sanqpo -* samppo ‘stroll, walk’, han pen -* ham pen ‘one time’, cwunpi -*■ cwumpi
‘preparation’, sinmun -*■ simmun ‘newspaper’, (mos mek.e -*■ mot mek.e -*) mon mek.e -*
mom mek.e ‘can’t eat’; chinkwu -*■ chingkwu ‘friend’, pankawe -* pangkawe ‘is happy’,
sonq-kalak -* song-kkalak ‘finger’.
In faster speech, m is replaced by ng before k:
camqkan -*■ cangkkan ‘a while’, nemkye -* nengkye ‘across, over’, Imkum -*■ Ingkum ‘king’,
cikum kkaci -* cikung kkaci ‘up to now’.
Sometimes, in fast or sloppy speech, pk(k) is replaced by kk:
komapkeyss. s up.nita -* komakkeyssumnita ‘I will be grateful (to you)’, poypkeyss. s up.nita -*■
poykkeyssumnita ‘I will see you’.
In sloppy speech ng often drops between vowels, especially when it is before i or y:
cwungang -» cwuang ‘central’; congi -*■ coi ‘paper’; tongyang -*■ to.yang ‘Orient’; the
placenames Phyengyang -* Phye.yang (-* Pheyyang, §2.7.9), Yangyang -*■ Ya.yang; .. .
Sometimes a final -ng is dropped: the hapsung (jitney) boys in 1960 Seoul would call out slche! for
slcheng ‘City Hall’. In those days when you left a Seoul restaurant you might hear a cordial A(n)nye
kapsye! for Annyeng hi kapsio! ‘Good-bye, sir’.
A casual reduction of m to w can be heard in rapid versions of ku man twue (yo) > kuwantw[u]e(yo)
(§2.7.7) > k[u]wante(yo) = kwante(yo) > k[w]ante(yo) (§2.7.4) > kante(yo) ‘cut it out (= stop);
let it go (at that)’.
2.7.2. Vowel length variants.
Vowel length is distinctive in Korean, and the long vowel can be considered as a string of two
identical short vowels. But many speakers do not use long vowels in all the words for which some
speakers retain them, so that most words with a long vowel within a morph have short variants. Even
for a speaker making maximum use of the length distinctions, the long vowel is usually restricted to
the first syllable of a word, so that virtually every morph with a long vowel has a grammatically
conditioned alternant with the short counterpart, as can be seen from the pair of synonyms ungpo =
poung ‘retribution’. But not all short-vowel morphs have long alternants or variants; many are always
short. For practical purposes, I indicate a variable long vowel - any long vowel within a morph - by
putting a macron above the letter symbol. In some Korean dictionaries, a long mark is put over the
entire syllable; in others, the syllable is followed by two dots, like the colon that is used to mark vowel
A Reference Grammar of Korean
PART I 33
length in the International Phonetic Alphabet. A problem arises as to whether the length should be
marked LEXICALLY, and accordingly seldom written on a syllable that is not initial in a word, or
MORPHEMICALLY. The synonym pair cited above showed a lexical marking of the length; a morphemic
marking would be ungpo = poung. South Korean dictionaries have generally shown the length
morphemically (and etymologically) for words of Chinese origin. The North Korean dictionary (NKd),
and the Yale dictionary (KEd), mark the length lexically, as is the general practice of this book, though
in a number of examples the morphemic marking will be seen. Lexical marking seems to be the safer
approach, since lexical units can be readily checked with native speakers, and many problems arise in
morpheme identifications and our decisions on “basic” length for both Chinese and non-Chinese
morphemes. The Chinese characters are not always a reliable guide: the cities of Taykwu and Taycen
are pronounced with short first syllables that represent the character tay ‘large, great’, for which the
length may be heard in words like tay-hak.kyo ‘university’.
The words kiil ‘fixed date, term’ and kll(q) ‘to be long’ at times are homonyms for some
speakers; at other times for those speakers and at all times, perhaps, for other speakers, they are not.
In rapid speech both words may even sound like kil ‘road’. The word chwiim ‘inauguration’ is to be
pronounced in two syllables, with the first long, but it is often said as a single long syllable, and in
rapid speech even that may be heard shortened. The distinction between e and e is the easiest to hear,
for in Seoul speech the long variety of that vowel is conspicuously higher in quality, and the short
variety is not only lower but backer and more rounded. That helps distinguish heen ‘falsehood’ from
hen ••• ‘old, worn(-out)’, and tepta ‘is hot’ from te epta ‘shoulders more’ (An Sangchel 1988:120) and
te eps.ta ‘lacks more’. For other vowels the length is mainly observed when needed to distinguish
particular sound-alikes in certain contexts. For more examples of vowel length, see KM 8-9.
When a one-syllable phrase ends in a vowel, that vowel is automatically lengthened, so that when
# cited in isolation the words si ‘poem’ and si ‘city’ are identically /sii/. The automatic lengthening
disappears when the word is part of a longer phrase: si to ‘poem too’, si to ‘city too’. This kind of
lengthening is ignored by all the orthographies and most of the linguistic descriptions. (A similar
phenomenon is common in western Japan and in the Ryukyus.)
He Wung (1965:89) points out some length alternations in verb forms. The vowel length found in
many of the verbs here called 1-extending, s-dropping, and -w- (p-leniting) stems is lost before endings
that begin with a vowel - ke-1- -» kel.e ‘hangs’, cl(s)- -* cie ‘builds’, kow- -* kopta but kowa ‘is
pretty’ - and also in voice-derived forms such as alii- ‘informs’ *- a-1-. The stem pe-1- ‘earn money’
is an exception (pel.i ‘earning money’, pel.e ‘earns money’), probably because it is a reduction of
peu-1-; similar exceptions are kku-1- ‘pull’ (kkul.e), coming from earlier kuu-1-, and sse-1- ‘chop’
(ssel.e), which goes back to sehul- (LCT 449b). With the exception of tut- ‘hear’, the verbs that end in
leniting t (the verbs here treated as modern 1-final stems = -T/L- stems) have vowel length before a
consonant, where they preserve the original stop unlenited: ket.ko ‘walking’, kel.e ‘walks’. The length
on these stems resulted from the blend of a Middle Korean low tone on the first vowel and a high tone
on a lost vowel that must have followed the consonant before the lenition took place: *ke 'tu- 'kwo >
"ket- 'kwo, then *ke 'tu- 'e > ke 'l-e. A similar history accounts for the vowel length of s-dropping and
-w- stems (Martin 1973; Ramsey 1975, 1978); see below. On the accentual exception of tut- ‘hear’ it
is interesting that this seems to be the only one of these stems that Phyengan preserves unlenited, with
the infinitive tut.e = Seoul tul.e despite Seoul-like sil.e ‘load’ and kel.e ‘walk’ (Kim Yengpay
1984:53).
The vowel length of a Chinese morpheme usually drops when it is noninitial in a word; this is part
of a general tendency to retain accentual distinctions only in the initial position.
In addition to the sort of lexical length mentioned above, there also occurs an “expressive
lengthening” as a voice qualifier, often accompanied by rasp or other voice qualifiers, for certain
mimetic words, such as the last syllable of adjectival nouns ending in -us (hata). Cf §14.
Younger Seoul speakers have largely lost the old vowel-length distinctions but they have new long
vowels based on dropped —h- (§2.7.3) or -u••• (§2.7.5), or in words borrowed from English and
34 PART I
A Reference Grammar of Korean
other modern languages. The older vowel-length distinctions are ignored in Hankul orthography. The
new long vowels in modern loanwords are also ignored in the North Korean orthography but in the
south the length is sometimes written with a repeat of the vowel as a separate syllable (with the zero
initial), though occasionally you will see instead a dash (-), a usage borrowed from the Japanese
treatment of katakana words. You may find khullm ‘cream’ spelled “khu-lim”, “khu-li-im”, or even
“khu-li—m”.
Decisions on noting vowel length for certain common words can be troublesome. We have
followed LHS, KEd, and NKd in writing the stems clna- ‘go past’ and clnay- ‘go past it’ (and their
derivatives) with a long first vowel, and that is historically correct, but Kim Minswu follows
contemporary Seoul standards in writing them short, and the student is advised to treat them as short
despite our retention of the length. We write the noun sihem ‘examination’ without vowel length, but
dictionaries list it as sihem (KEd, NKd) or sihem (LHS), and the usual pronunciation today is /syem/,
with the intervocalic -h- dropped (as usual) and the i losing its syllabicity (reduced to y) but with
compensatory lengthening of the following e. The word traditionally spelled iyaki ‘talk, tale’ is usually
pronounced yayki or (more commonly) yeyki, and all three versions will be found in this book. The
first vowel of the stem komaw- ‘be grateful’ is short, not long, but sometimes “expressive” length is
superimposed in saying Ko(:)mapsup.nita, and that is responsible for the misleading remark in KEd
141b “Some pronounce long [komapta]”. There are probably a few other cases of this sort that have
escaped my attention. The student need not worry about vowel length except when he hears it, for
younger Koreans pronounce most of the older long vowels as short, maintaining only those long
vowels that are the result of contraction, such as mam < maum ‘soul’, and the newly arrived long
vowels that have come in with modern loanwords (sometimes written double, as geminates): khatu =
khaatu ‘card’, aphathu = aphaathu ‘apartment house’, slcun = siicun ‘season’, eythosu =%
eyeythosu ‘ethos’, phesuthu = pheesuthu ‘first (base)’, khotu = khootu ‘cord’, khyu = khyuwu
‘cue, Q’. Sometimes a bar is written for the second vowel, and sometimes it is simply ignored. The
diphthongized English long vowels are usually treated as eyi and owu.
The distinctive length of the central area corresponds to distinctive pitch or a combination of pitch
and length in certain other parts of Korea. In the province of South Kyengsang, for which we have He
Wung’s description of his native dialect of Kimhay, there are three lexically distinct pitch levels HIGH,
mid, and low. (The high pitch sometimes has a slight fall, especially on a monosyllable in isolation.)
In North Kyengsang (e.g. Antong) and also in North Cenla (e.g. Kwunsan), there are only two
lexically distinct pitches high and LOW, and part of the distinction is carried by vowel length. The low
pitch of Kimhay is cognate with long low nuclei in Antong and the mid pitch of Kimhay is cognate
with SHORT low nuclei. (Apparently there are no long vowels in Antong with HIGH pitch.) Farther
north, in Hamkyeng (e.g. Hamhung and Hoylyeng) high and LOW pitch are distinguished but there is
no cognate distinction of length. Moreover, both the mid and low pitches of Kimhay are cognate with
the HIGH pitch of Hamhung, and the LOW pitch of Hamhung is cognate with the high pitch of Kimhay.
In each of the Korean dialects the situation is complicated by a certain amount of pitch sandhi (partly
described by He Wung) that is similar to the length alternations of standard Korean. We refer here to
the “basic” accents of words. The distinctive lexical pitch is not to be confused with the sentence
intonation of standard Korean, described in §2.8. Speakers from Seoul and from most of the north
and west do not differentiate words by pitch alone. But many speakers from the south and east retain
their native distinctions of pitch even after they have adjusted their pronunciation to the standard
language quite well in other respects.
Below is a table that shows a few stock examples to demonstrate the cognate relationship of pitch
and length in the dialects. The first column lists the examples in standard forms; the other columns
show the pitch and length for each example in the four dialect types known to me.
A Reference Grammar of Korean
PART I 35
Standard Seoul
Kimhay
Antong-Kwunsan
Hamhung-Hoylyeng
mal ‘horse’ short
pay ‘pear’
son ‘guest’
high
high
low
mal ‘measure’ short
pay ‘stomach; boat’
son ‘hand’
mid
low short
high
mal ‘words’ long
pay ‘double’
son ‘loss’
low
low long
high
In Middle Korean the syllables of the second type were preceded by a dot, representing high
pitch, and those of the third type by a double dot, which represented a long rise going from low to
high. In the Yale Romanization the two accent marks of Middle Korean can be represented with a
raised dot and a dieresis (raised double-dot): swon ‘guest’, swon ‘hand’, "swon ‘loss’.
From a typological point of view, it can be said that Korean words have lexical accent,
manifested by pitch or length or a combination of pitch and length, depending on the dialect.
Somewhat similar remarks can be made about Japanese. But the tones of Chinese are different: they
represent a pitch contour that is part of each monosyllable, just like the consonants and vowels. The
accent patterns of Korean and of Japanese spread over longer stretches, since these languages have
many polysyllabic words and morphemes.
2.7.3. Disappearing h.
The phoneme h freely drops between typically voiced phonemes (the vowels and y w m n ng I):
a[h]op ‘nine’, slm-[h]i ‘extremely’, man[h].i ‘lots’, sin[h]on ‘new marriage, honeymoon’, un[h]ayng
‘bank’, cen-[h]ye ‘totally’, kyel[h]on ‘marriage’, chel[h]ak-cek ulo ‘philosophically’, pang[h]ak
‘school vacation’, annyeng [h]i ‘in good health’, ko[h]yang ‘hometown’, kyo[h]wan ‘exchange’,
sel[h]wa ‘story, tale’, in[h]yeng ‘doll’, um[h]yang ‘sound, noise’. The h-less version of ko(h)yang i
‘home town [as subject]’ can sound like ko.yangi ‘cat’, though in Seoul that word usually contracts to
kwayngi. And ol hay ‘this year’ can sound like olay ‘for a long time’ (both /oley/). There can be more
than one dropped h in a phrase: 'lhay hay (yo) ‘I understand’ is often reduced to /Ieyey(yo)/.
For certain words the version with the elided ™h™ is now considered standard: puengi ‘owl’
(dialect puhengi, puhii) < 'pwuhwe'ngi, hwang’a = hwanghwa ‘sundries, variety goods’ < hwang-
'hwa (for the dropped w in these two words, see §2.7.4); pinye ‘hairpin’ < pinhye (? < PIN ‘hair on
temple’ + hye ‘tongue’ or °hye- ‘pull’); ili ‘wolf’ < ilhi. From a strictly synchronic point of view,
there is no h in words like man(h).i, despite the spelling, which is morphophonemic; we infer the h
from other forms such as mankho *- manh.ko, as it is spelled. In unfamiliar Chinese words, the
underlying h will emerge as a kind of reading pronunciation, but it usually drops when the word comes
to be often said. Inflected forms of h-final stems are pronounced without the h when it is between
voiced sounds, but occasionally you will hear the h restored for emphasis in certain forms, as in
Coh.un kyey[h]oyk ita ‘It is a GOOD plan’, though never in other forms, such as co[h].a > /cowa/,
which can be emphasized only by further lengthening the vowel (co:wa), for the h has been completely
absorbed in the infinitive, and an epenthetic glide w has taken its place. How old is the h-elision and,
after a rounded vowel, the epenthetic w? 1882 Ross 35 has nwo.wa.la < noh.a la. There are examples
of n[h] in the 1500s and 1700s: "ma'ni ( ? 1517~ 'No 2:26a, Pak 1:20a) and man.i (1703 Sam-yek 5:2)
= man(h).i ‘much’; skun.e (1736 n Ye 3:13 [LCT]) = skunhe (1783 Cahyul 1) = kkun(h).e ‘end’.
Sometimes the entire h ■■ syllable drops: Na [ha]nthey? ‘For me?’; Kulena silphay [hay]ss.c[i]
yo ‘But they failed’. The verb expression in Pusan ey se Kim Sencwu thukpha-[w]en i poto hap.nita
‘From Pusan, correspondent Kim Sencwu reports’ is often said as poto [ha]p.nita = /potomnita/ and
equally often as pot[o h]ap.nita = /potamnita/ < /pot[w]amnita/ (o > w, §2.7.7). A similar
example: kac[i] an[h].ulye ko hap.nita = /kacanulyekamnita/ < /••• k[w]amnita/ ‘I won’t go’.
36 PART I
A Reference Grammar of Korean
2.7.4. Disappearing w.
Before a mid or low vowel the phoneme w freely drops after p, ph, ps, m, wu, or o: sam-[w]el
‘March’, sam [w]en ‘three wen [monetary unit]’, kwu [w]en ‘nine wen’, kyo[w]en ‘teacher’, o-[w]el
‘May’, cem.[w]en ‘shop clerk’, pep.[w]ang ‘pope’, m[w]e ‘what’, cip.h[w]ey = cip.hoy ‘meeting’,
cham.[w]ey (-* chami -*■ chaymi, §2.7.9) = cham.oy ‘melon’, ip.[w]en ‘entering hospital’,
caps[w]e ‘partakes’ (< capswue, §2.7.9). The usual way to say Mai hay pwa ‘Tell me about it’ is
[mareba] = Mai [h]ay p[w]a. There are diachronic examples of postconsonantal w dropping even from
wo, which is usually taken to be a monophthong vowel: pacwo (1562) > paco (1748) > paca ‘reed
fence’.
In sloppy speech (and widely in Seoul) w often disappears after nonlabial sounds, too, when a mid
or low vowel follows: si[w]en hata ‘is cool’, an [w]ass.ta ‘didn’t come’, towa c[w]e ‘does the favor
of helping’ (< cwue, §2.7.7), n[w]a t[w]ess.e ‘put it away’ (< noa, §2.7.7) < no[h].a twuess.e,
tongmul-[w]en ‘zoo’, chil-[w]el ‘July’, ceng-[w]el ‘January’, pyeng[w]en ‘hospital’, kong[w]en
‘park’, ceng[w]en ey ‘in the garden’, tay[h]ak-[w]en ‘graduate school’, cik.[w]en ‘staff, personnel’,
k[w]ank[w]ang-kayk ‘tourist’, meych [w]el iess.na yo /meyt[w]eli(y)enna.yo/ ‘what month was it?’,
kh[w]aysok hata ‘is speedy’ (the first syllable is typically devoiced), h[w]ankap ‘60th birthday’,
n yen[hw]ey = "yenhoy ‘annual meeting’, chen [hw]an ‘a thousand hwan (outdated monetary unit =
wen)’. As the last two examples show, hw can drop as a string: pho[hw]an ‘cannonball, shot (put)’,
so[hw]al [h]i ‘sloppily, carelessly’ (also /sowali/, /swali/). For many speakers the phrases cen[hw]a
’ta ‘it’s a phone call’ and cen [h]ata ‘reports’ can converge. The string hw, when not dropped
internally or when initial, is articulated by many speakers as a bilabial fricative [F], as noted in §2.4.
From the viewpoint of our Romanization there is an interesting case in tewun - ‘ - that is warm’
-*■ teun - -* (§2.7.5) ten ••• . Notice that all cases of -owun, -owul, and -owum freely contract to
-on, -61, and -om; and all cases of -wuwun, -wuwul, and -wuwum to -wun, -wfll, and -wum.
Because oy is generally pronounced as wey, words spelled with that diphthong often end up just
as [w]ey, so you may hear an tey = an t[w]ey for an toy(e) ‘it won’t do; too bad’. Other examples:
pemc[w]ey = pemcoy ‘crime’, sath[w]ey = sathoy ‘declining/refusing office’.When you hear
/keyley/ it may escape you that this could be k[w]eyl[w]ey and the word can be found in the dictionary
as koyloy ‘puppet’. The h of hwey (< hoy) freely drops along with the labial, but only when a voiced
sound precedes, as in wi[w]en[hw]ey = wiwen-hoy ‘committee’, and in rapid speech that can even be
compressed further to /weney/! On the other hand, um.ak-hoy ‘concert’ can be reduced to umakhey
but not *umakwey or *umakey (the latter could only represent um.ak ey ‘in/to/of the music’), and
kwuk.hoy = kwuk.hwey ‘national assembly’ can shorten to kwukh[w]ey = kwukhey but not to
*kwukwey or *kwukey. For thoywen hayss.ta ‘got out of the hospital’ you will hear (thweywen >
th[w]ey[w]en =) /theyen[h]etta/ or even just /theynetta/.
Further compressions may baffle the ear. When i or u are left directly preceding a vowel by the
eliding of h or w (or hw), the high vowel often loses its syllabicity and becomes a semivowel glide:
si[h]em ‘test’ > syem with compensatory lengthening of the remaining syllable. And swuep > swep
(§2.7.7) ‘class instruction’ may be further reduced to sep. In rapid speech you will hear drastic
reductions such as kyeyhoyk = kyeyhweyk > kyey[h]weyk > kyey[w]eyk > kyeyk = /keyk/
‘plan’. The expression swlpsseykoki is from sw[u]ipss[w]eykoki = swuip soy-koki ‘imported beef’,
and in dialects you may hear swlpsseykeyki < -k[w]eyki < -koyki < -koki with partial
assimilation (fronting or “umlaut”) of the next-to-last vowel to the final high front vowel. The form
swlp does not further reduce to s[w]Ip because the w drops only before a mid or low vowel, not i or u.
But notice that some instances of hwu reduced to hw (§2.7.7) are followed by nonhigh vowels, and w
before a mid or low vowel freely drops, so that in rapid speech (ku) hwu ey ‘afterwards’ > (ku)
hw[u]ey = (ku) hwey sounds like (ku) hwey = (ku) hoy ‘(that) meeting’ and both can be further
compressed to (ku) h[w]ey = /(ku) hey/. The city of Swuwen is often called /swen/. The word
kan[h]o-[w]en ‘nurse’ will drop the h and/or the w, and /kan(h)oen/ can be further compressed to
/kanwen/, which in turn may drop its w leaving the listener with /kanen/ to puzzle out.
A Reference Grammar of Korean
PART I 37
2.7.5. Postvocalic u.
Sequences of vowel + u are often pronounced with vowel length replacing the u: kium, klm
‘weed’ (whence klm ‘seaweed’?); taum, tarn ‘next’; maum, mam ‘soul, heart, mind’; cheum, chem
‘for the first time’; maul, mal ‘village’; keyuluta, keyluta ‘is lazy’; koul, kol ‘district, county’. Since
h drops readily a sequence of vowel + h + u is often reduced to a long vowel: noh.una, nouna, nona
‘puts it but’, tah.uni, tauni, tani ‘since it arrives’, coh.umyen, coumyen, comyen ‘if it’s good’. Cf
§8.1.2, §8.2.4.
A similar reduction whereby i (after Vy) behaves as u does above will be heard in nayl *- nayl *~
l nayil ‘tomorrow’. (The suppression of the newly acquired vowel length remains to be explained.)
2.7.6. Intercalated semivowels.
The vowel component FRONT occurs in the phonemes i ey ay (oy) (wi) and the component
rounded in wu o (oy) (wi). These two features freely overlap a following vowel to spawn an
etymologically unmotivated semivowel y or w. From pi ‘rain’ + os ‘garment’ comes pi os ‘rain-gear’,
which sounds like /piyot/, The infinitives chwue ‘dances’ and chwuwe ‘is cold’ ordinarily sound the
same /chwuwe/, and the infinitives peyye ‘gets cut’ («- peyi-, a passive stem) and peye ‘cuts’ (•*-
pey-, a transitive stem) are often indistinguishable as /peyye/. Some speakers try to differentiate words
like kiyak ‘weakness of spirit’ and cwuwi ‘surroundings’ from the quasi-homonyms kiak ‘instrumental
music’ and cwuuy /cwui/ ‘ism’ by holding on to the y and w. Other speakers, however, distinguish a y
or w which is motivated (morphophonemically expected from our knowledge of other alternants, e.g.
in yakca ‘weakling’ and sawi ‘all around’) from one which is simply the predictable lag in phase of a
phoneme feature. Such speakers make a difference in pronunciation between nayo ‘puts out
[authoritative style]’ (< nay-o) and nay yo ‘puts out [polite style]’ (< nay-e yo). An analogous
situation occurs in English with words like “prints” and “prince”, which are seldom if ever
distinguished in speech.
The practice of the Korean Language Society is to write y or w in all cases of semivowel except
when there is a clearcut etymological reason not to do so. Within a morpheme the semivowel is
written: iyaki ‘story’ (contrast i aki ‘this child’), kwiyal ‘paint-brush’, swuwel hata ‘is easy, handy’.
The two apparent exceptions of ppay-as- ‘grasp’ and payam ‘snake’ either show the influence of their
abbreviations ppays- and paym or else reflect an etymological analysis (Cf as- ‘snatch’, dialect
variants piam and pi-emi ‘snake’). Other apparent exceptions are kayam ‘hazel nut’ and sayang
(abbreviation sayng) ‘ginger’ - but sayyang also occurs.
The phenomenon extends to cases of wu or o + the disappearing h of §2.7.3: no(h.)a ‘puts’ and
towa ‘helps’ rime for most speakers, and both are sometimes shortened to nwa and twa (§2.7.7).
These phenomena have been attested for some time: nwowala (1882 Ross 35) = noh.a la ‘put it
[there]!’ The intercalated palatal semivowel can be seen in the Middle Korean spelling of -i/y +
particle 'ey as “-Wy'yey” (in contrast with the “-Wyey” that represents -WyGey).
2.7.7. Desyllabiflcation of i, wu, and o.
Sequences of i + vowel or of i + y + vowel are often reduced to y + vowel. The vowel is
usually lengthened if it is in the first syllable after pause. The most conspicuous examples are of the
infinitives of stems that end in i: kitalie -* kitalye, masie -» masye (usually pronounced mase, at
least by older Seoul speakers), kacie -*■ kacye (usually pronounced kace), kaluchie -* kaluchye
(usually pronounced kaluche), titie -» titye, ... . Since the honorific marker is -(u)si-, the sequence
-(u)sye is especially common - and usually pronounced as if -se, at least by older speakers, but among
the younger generation the pronunciation with /sy/ [S] is prevalent and it seems to be spreading.
Example: kasie -*■ kasye (-» kase). Shortening of longer infinitives in -ie is standard practice in
written Korean nowadays. The one-syllable stems are usually not abbreviated in writing but in speech
you hear the same sort of shortening, usually with compensatory lengthening of the vowel: ttye for ttie
‘wears a belt’, phye for phie ‘smokes (= phiwe); blooms’, chye (pronounced che) for chie ‘hits’, eye
(pronounced ce) for cie from either cl(s)- ‘build’ or ci- ‘bear on one’s back; ... ’.
38 PART I
A Reference Grammar of Korean
The vowels wu and o are often reduced to w before a vowel and to nothing before w + vowel,
especially when the result is not immediately followed by a pause: ewe for cwue ‘gives’, twe for twue
‘puts away’, twa for towa ‘helps’, mwe (further reduced to me by §2.7.4) for mue ‘what’. The
phenomenon extends to words in which the h disappears (§2.7.3): nwa for noh.a ‘puts’, ewa for coh.a
‘is good’. The vowel sometimes lengthens: twa yo (= towa yo), nwa yo (= noh.a yo), ewa yo (=
coh.a yo). The reduction of longer infinitives in -wue is in written Korean nowadays but the one-
syllable stems ending in wu are not usually abbreviated in writing, except that cwue and twue as
auxiliaries are often written ewe and twe. The -w- then often drops in rapid speech (§2.7.4).
In general the vowel of monosyllabic infinitives reduced from i-e, wu-e, and (w)o-a are basically
long (eye, ewe, twe, nwa, pwa, wa) but the length is suppressed in the past forms: compare wa iss.ta
‘is (come) here’ and wass.ta ‘came, has come’. And the length is often dropped when the infinitive
closely follows other another form, as the auxiliaries often do. In this book we follow LHS (and we
correct KEd) in writing chye ’ta (pota/poita) and chye tulta despite the seeming irregularity of chye
cwuta/kata/nayta/pelita. NKd has chye ’ta pota but chye ’ta poita and, I believe, chye tulta (the
photoprint is unclear). The long vowel in the infinitive wa ‘come’ is questionable; the contraction was
the usual form in Middle Korean and marked with a single dot, not the double dot that would lead to
the modern length. If genuine, the length may be new.
The short vowel of the infinitive in phye cita owes to the earlier (and dialect?) form phyeta =
modern Seoul phita. The infinitive khye < khyeta = khita is similar. These infinitives, like that of
literary (and dialect?) seta = suta ‘stand’, simply absorb the infinitive ending -e with no compensatory
lengthening, just as ka < kata absorbs the -a. A stem that ends in unrounded u drops that vowel when
it attaches the ending, and there is no lengthening of the resulting syllable: khe < khuta, sse < ssuta,
tte < ttuta, the < thuta. The irregular length of kke cita, the cita (and for LHS and Kim Minswu
the ttulita) is anomalous. Also anomalous is the length of kke from kkuta, for which no good
explanation is apparent; the earlier form was pske (1462 ’Nung 2:43b, 1481 Twusi 25:13a) with but a
single dot, and the earlier form of the was similar, 'pthe (1481 Twusi 7:24b). (LHS lists kke without a
long mark, thus short, but all of his compound verbs have kke. NKd has a long mark for kke itself and
a number of the compounds, but strangely leaves kke cita unmarked.) There are several pairs of
expressions that show irregularity with respect to vowel length. We hear, for example, kkay cita, kke
cita, and the cita but kkay ttulita, kke ttulita, and (according to NKd and KEd) the ttulita, though
LHS and Kim Minswu have the ttulita, as if the geminate tt had curtailed the length. NKd strangely
has short ph’e cita despite ph’e ttulita; both are long in the other sources.
Other cases of vowel reduction are often heard in the casual construction -ci yo -*• -c[y]o, in the
command form -usio/-usipsio -*■ -us(y)o /-usips(y)o, and in an occasional noun, such as kyek ca for
kiek ca ‘the letter K’. I have heard /aneyyo/ = an’ ey yo *- anyey yo *- ani (y)ey yo ‘it isn’t’ *- ani
+ ie(y) yo. Also: Wyil -*■ 'nayl -*■ *nayl (§2.7.2) ‘tomorrow’; (§3.5.[9]) oylye *- oilye (§2.7.3) <
ohilye (< wohi'lye ) ‘rather’; toylye *- tolie (< tolihye < twolo-'(h)hye) ‘conversely’.
2.7.8. Reduction of wie.
The sequence wie is often replaced by oy (as if by way of *wye) and oy is often further replaced
by wey, as noted earlier. Examples are swle soy ‘sours’, swless.e soysse ‘soured’; ttwie ttoy
‘jumps’, ttwiess.e -*■ toyss.e ‘jumped’. The standard Hankul spelling writes this reduction as “wey”,
apparently influenced by the many speakers who do not distinguish oy from wey; compare the spelling
of “way” for the infinitives of stems ending in oy, §9.4. The stem sakwi- (sakwie sakoy) is an
irregular development from sa'kwoy- ‘get acquainted’.
2.7.9. Vowel assimilation.
The vowel ey is frequently replaced by i in rapid speech when the following syllable contains i or
y: ciil -» cey-il ‘number one’, kitali yo *- kital(y)ey yo *- kitalye yo ‘waits’, kitali ya *- kital(y)ey
ya kitalye ya ‘only by waiting’, cikhi ya ’nta *- cikh(y)ey ya ’nta *- cikhye ya hanta ‘must
maintain’; kasi yo *- kasey yo *- kasye(y) yo ‘(someone esteemed) goes’; - i yo *- ••• (y)ey yo *-
A Reference Grammar of Korean
PART I 39
••• ye yo *- - ie yo ‘it is’; hwuli chinta *- hwuley chinta *- hwulye chinta ‘lashes, whips’.
We also find pairs of words in which one member, usually the more common form, has a front
vowel either after c(h) or before a syllable that contains i or y. Examples: achim, achum ‘morning’;
ilccik(-i), ilccuk(-i) ‘early’; ayki, aki ‘child’; teyli-, tayli-, tali- ‘take (someone) along’; kitayli-,
kitali- ‘wait’; yeyki, yayki, iyaki ‘story, talk’; tay(n)ni-, ta(y)ngki-, ta(n)ni- ‘go back and forth
regularly’; caymi, cami ‘fun’; hayk.kyo, hak.kyo; teyngi, tengi ‘lump’; hayngkil ( *- hangkil
§2.7.1) +- han-kil ‘ street’. Notice also meychil ‘how many days’ < myech-(h)ul, not to be
misinterpreted as containing the Chinese il < 'QILQ ‘day’. More complicated explanations are needed
to account for taynchu = tanchwu ‘button’, weynsswu = wensswu ‘enemy’, mayntunta =
mantunta ‘makes’, oynthong (or weynthong) = on-thong ‘entirely’. The adjectival noun weyn-man
hata ‘is fairly good’ is a reduction of wuyen-man hata. Where it causes no confusion the assimilated
form has been standardized as the spelling for some of the words. One of the vexing problems is with a
large group of voice-derived verbs (causatives and passives) in which a vowel of the stem is often
assimilated, and there are back formations “correcting” a legitimate lil to /u/ (or, after a labial, /wu/).
In question are words like cwuk.i- ‘kill’ (often pronounced cwiki-) from cwuk- ‘die’, mek.i- (often
meyki-) from mek- ‘eat’, sok.i- ‘cheat’ (often soyki-) from sok- ‘be cheated’. In the appropriate part of
Appendix 1 there is a comprehensive list with cross references from the spoken assimilations or back
formations to the standard written forms.
The word soycwu *- socwu < "sywow-"cywuw ‘hard liquor’ may reflect metathesis of the glide
and/or assimilation. But soy-koki < "sywoy-kwo ki ‘beef is contracted from * sywo oy kwo'ki = so
uy koki ‘meat of the ox’ (Cf LCT 1971:223; the accentual anomaly is unexplained). For the compound
talk ‘chicken’ + al ‘egg’ the expected pronunciation would be /takal/, but instead the standard written
form is talk yal (spelled phonemically tal-kyal), and that seems to come from talk [u]y al /talk[e]yal/
< tol'k oy al ‘egg of chicken’. Common variants include talk eyl, talk ayl, and talk yayl. (The
possibility of a form like *talk yeyl is excluded by §4.3.) Perhaps similar is the pronunciation /silye/
for /sile/ silh.e ‘I dislike if, which is popular today among young women in Seoul, but seems to have
been around for a while (1936 Roth 185 gives silh.e/silh.ye). Yet no other “Ih- has that sort of
variant: /kkulye/ means only kkulh.i-e ‘boil if and kkulh.e ‘it boils’ is pronounced /kkule/. It might
be thought that the intruded palatal glide of /silye/ is due to the i of the preceding syllable, but there is
an attestation of sulhye in 1887 Scott 63, apparently made on the earlier version of the stem, which
was sulh-: kaki sultha ‘I don’t care to go’ (1887 Scott 80). One explanation might be a shortening of
sul ho.ye > sil-h[]ye, with the irregular infinitive ho.yalho.ye of °ho- > ha- ‘do’, for which the
attestations of sul-ho.ye (1676 Sin.e 9:10b), sul-'hu'ye (1586 Sohak 5:9b), and earlier 'sul-'ho'ya
(1447 Sek 13:18a) provide support. Compare modern cen-hye ‘entirely’ < cen + ha.ye and hayng-
’ye < hayng-hye ‘by chance’ < hayng < HHOYNG + ha.ye.
Less commonly i is substituted for u in attaching endings to stems that end in -s-, -ss-, -c-, or
•ch-: wusina = wus.una ‘laughs but’, issina = iss.una ‘there is but’, chacina = chac.una ‘finbuf,
cochina = coch.una ‘follows it but’. The popular Seoul pronunciation til.ye ’ta ponta for tul.ye ’ta
ponta ‘peers into; looks (gazes) at’ assimilates the first vowel to the following palatal syllable.
A common phenomenon, especially in the north, is the reversal of ye to ey: pey for pye ‘rice plant’,
making it a homonym of pey ‘cloth’ and also of the pey which is a variant of pay ‘boat; stomach’ for
those who do not make the ay + ey distinction; Phey(ng)yang for Phyengyang (name of city - often
spelled “P’yang” in the headlines of English-language newspapers in Korea); peyng(w)en <
pyengwen ‘hospital’. The written word myech ‘how much’ is usually pronounced meych even in
Seoul, and it is so written here.
2.8. Standardization variants.
A number of words appear in several shapes, either phonemically or just orthographically, and
they reflect different notions of what is “standard” Korean. In some cases, the words are isolated
instances, in other cases they reflect more general problems. There is considerable agreement among
the Korean grammarians, in both the north and the south, on most of the isolated cases and on many of
40 PART I
A Reference Grammar of Korean
the general problems. Where my own observations of current standard usage agree with the decisions
of the Korean grammarians I simply use their spellings without comment. In other cases, I have spelled
out my differences of opinion, as in my preference for tat.e (etc.) over tat.a as the infinitive of tat-
and of hay over ha.ye as the infinitive of ha- (§1.9, §9.3) and my preference for -tuni over -teni as
the retrospective sequential (see the entries in Part II). I have been somewhat crankier than the Korean
grammarians in insisting on distinctions between iya (particle) and ia = ie (copula infinitive), between
iyo (particle) and io (copula); with most of them, I deplore the writing of intercalated “y” within forms
of the copula (“iye” for ie etc.). But I appreciate the difficulty faced in making these decisions and
recognize that most people prefer to write the intercalated y within inflected forms without worrying
about the internal structure, since the contraction of (—)ie is (-)ye. In general, I frown upon the
widespread writing of words and phrases in abbreviated forms, since that obscures the grammar and
often leads to confusion. In this book I have used an apostrophe to indicate omitted letters, as in ’— =
i- or ha-. In some cases, current South Korean orthographic practice is at variance with the North
Korean and those differences I have shown by superscript letters, as follows:
This book
North Korea
South Korea
1 . iss. s up.nita ‘is’
iss.sup.nita 1
iss.up.nita 2
iss. s o ‘is’
iss.so 3
iss.o
2 . an h ay ‘wife’
anhay
an ay
3 ol h -paluta ‘is upright,... ’
olh-paluta 4
ol-paluta
4 . u y lyey ‘usually’
uylyey
ulyey
5 . m y ey [Chinese morphs]
mey
myey
ph y ey [Chinese morphs]
phey
phyey
(See §4.3; also §§1.5-6.)
1 An example appears in Mkk 1960:4:26 [sic],
2 But the 1988 revised rules of the Ministry of Education abolishes the spelling “-up.nita”
and writes -sup.nita whenever a consonant precedes. Presumably - s o is to be treated
similarly; the published rules neglect to inform us of that.
3 Cf Mkk 1960:3:26. 4 Pronounced /olpalu-/ according to NKd.
Earlier I had included here “mop s si” ‘very’, a word that has been spelled both mopssi and mopsi in
South Korea but only mopsi in North Korea, which disregards the etymology: mos < "mwot +
-(p)ssi < * 'psi, derived adverb < (-p)ssu- < psu- ‘use’; Cf mopssul ‘useless, no good’ from the
prospective modifier. But since the South Korean linguistic authorities, too, mostly favor ignoring the
etymology, we will write mopsi and treat the adverb as opaque. ('Yi Ungpayk 1961 gives the first
vowel as long for both mopsi and mopssul, reflecting the etymology, but that length is not reported by
other sources.)
The past -ess- and future -keyss- behave, of course, like iss- in group 1; many South Korean
grammarians agree with the North Korean spelling reflected in the superscript s , but in practice the
other spelling is more widespread in the south. We assume here that -sup.nita and -so are being used
after ALL consonant stems. If the less standard versions -up.nita and -uo are being used throughout,
then they should also be used after ss. That is, if you say mek.up.nita it makes sense to write
iss.up.nita, but if you say meksup.nita it would be more consistent to write iss.sup.nita. The
important thing is to use one or the other consistently (Cf §9.2). Notice also the remarks on
superscript 1 and n (§1.6).
There are a few words which, though historically -nn-, are actually pronounced -U~ (as if
coming from — n.l-). In South Korea the spellings have been standardized as -n.l-, but in North
Korea the historical spellings are used, despite the irregular pronunciation. We write -n.'n- :
A Reference Grammar of Korean
PART I 41
This book North Korea South Korea Pronunciation
kon.Wn ‘difficulty’ konnan konlan /kollan/
han.'nan-kyey ‘thermometer’ hannankyey hanlankyey /hallangkey/ (§4.3)
In parts of the north there are speakers who substitute II for nn in various words, such as allyeng for
annyeng. In Anpyen and Tek.wen of South Hamkyeng tullunta is used for /tunnunta/ < tut.nunta
‘hears’ (Kim Yengpay 1984:53).
There are other spellings that vary from the current spoken usage in Seoul:
1. The noun iyaki ‘story’ is usually said as yayki or yeyki.
2. The gerund -ko and the particle to are usually -kwu and twu, even in the speech of people who do
not substitute wu for o wholesale. (Cf §2.2.)
3. Less generally recognized is the substitution of e for a in many common words: - henthey for -
hanthey (particle), hekwu for hakwu for hako (particle), he(n)ta for ha(n)ta ‘does; is’.
4. The verb stem for ‘stand’ is written se- but pronounced su- even by Seoul speakers who do not
ordinarily substitute u for e (except in -tun for the written -ten, noted earlier). The verb stems
written phye- and khye- are pronounced phi- and khi- in Seoul; the stem written kenne- is
pronounced kennu-. The spellings in this book conform to the Seoul pronunciation.
5. An artificial spelling distinction: olun ‘right (in direction)’ olh.un ‘correct’. See §5.3.
I have followed the Korean grammarians in assuming only one standard treatment of the I-
doubling vowel stems (§8.3.1), such as pulu- ‘call’ and molu- ‘not know’ with their infinitives pulle
and molla, but many otherwise standard speakers double the I everywhere, pronouncing the stems as
pullu-, mollu-, etc. I have also followed the grammarians in the standard version of the intentive
-ulye/-lye, with a double 11 only when attached to the extended stem of the I-extending vowel verbs
(wullye ‘about to cry’ from wu-1-), but many speakers use a version with a double II everywhere:
-ullye / -llye. For such speakers we will have to say that the ending attaches, in the shape -llye, to the
UNextended stem of the I-extending vowel verbs. And many of those speakers use the vowel a instead
of e: -ullya/-llya.
There seems to be confusion among Korean grammarians over whether to spell -ulq ka as “-ulka”
or “-ulkka”. Some would like to treat anything that appears after a verb stem as an unanalyzable
ending, to be written phonemically. (Compare the remarks on -un ya in §1.9.)
For a fuller discussion of problems of standardization, see Martin 1968.
2.9. Intonations.
The following statements about intonation follow the analysis in Martin 1954 (= KM). Every
phrase or utterance of more than one syllable has a gradual nondistinctive rise throughout until the
onset of a particular intonation, which occurs near the end of the phrase and in conjunction with a
pause. The meaning of statement, question, suggestion, and command are sometimes carried (in whole
or in part) by the intonation, but often these meanings are wholly or partly expressed by morphs in the
ending of the verb form.
Seven intonations are recognized for Seoul speech:
1. PERIOD intonation (.): a fall, beginning on the third, second, or last syllable from the end of the
sentence (if on the last syllable this intonation is homophonous with 4).
2. COMMA intonation (,): a rise on the last syllable of a phrase.
3. QUESTION-MARK intonation (?): a rise on the third, second, or last syllable from the end of the
sentence (if on the last syllable the intonation is homophonous with 2).
4. exclamation-point intonation (!): a quick fall on the last syllable of the sentence, often
accompanied by a voice qualifier of overloudness.
5. double question-mark intonation (??): a dip on the third, second, or last syllable from the
end of the sentence (homophonous with 7 when on the last syllable).
42 PART I
A Reference Grammar of Korean
6. DOUBLE EXCLAMATION-POINT intonation (!!): a dip on the third, second, or (rarely) last syllable
followed by a fall on the next (rarely, the same) syllable.
7. triple-dot intonation (...): a dip on the last syllable of a phrase or sentence, often accompanied
by a voice qualifier of overlength.
Three phonetic features are involved: rise, fall, and dip. The dip can be described as a fall immediately
followed by a rise. King tells me children use - hanta as an exclamation with dip and rise (ha n ta).
The question-mark intonation primarily means QUESTION; the exclamation-point intonation shows
INSISTENCE. The meaning of the double question-mark is RHETORICAL QUESTION or lively, and that
of the double exclamation-point is lively and insistent. The comma intonation signals temporary
suspension and the triple-dot intonation expresses hesitation. The period intonation is the sentence-
final default when no other intonation is called for. The intonations marked ?? and !! seem to be
peculiar to Seoul speech and are largely limited to casual statements of the -ci (yo) type, but
occasionally occur with other sentences, as in Anl (yo)!! ‘No’. For examples of the intonations in
various types of sentence, see KM 62.
2.10. The earlier phonology.
Korean of the 15th-century Hankul texts, called (Late) Middle Korean and here dubbed “MK”,
differed from the language of the 20th century in offering a somewhat richer pattern of sounds and
strings of sounds. We can explore the earlier system through the Hankul spellings, making inferences
about the articulatory values of the written syllables and the environments where the same string of
phonemes was written in varying ways or where different strings were written as if they were the
same. From the dates of the texts carrying critical examples we deduce the relative timing of changes
in the articulatory habits of Korean speakers over a period of four or five centuries. From the patterns
found in the earliest Hankul texts we can surmise changes that must have taken place over the
preceding hundred years or so, and reconstruct patterns for that period with relative confidence.
Putting the accent patterns aside for separate study (§2.12), we will seek to date the changes in:
(1) the vowel system = syllable nuclei, (2) the initials = onsets, (3) the finals = codas, and (4) the
intersyllabic strings = interludes.
What clues we have to the pronunciation of Korean earlier than the 15th century, beyond what we
can obtain from internal reconstruction, are words written by means of Chinese characters, for the
most part intended as phonological representations of the Korean words. Interpreted in terms of the
MK pronunciations of the characters, the forms show few surprises, and there is little to suggest that
the system of sounds was drastically different from that recorded by the early users of Hankul.
2.10.1. The earlier vowels.
The earlier language had a vowel system similar to the modern system but with the addition of the
low back (and functionally unrounded) vowel o. In dialects other than that of Ceycwu the extra vowel
was lost and merged with other vowels. In noninitial syllables o merged with its higher counterpart,
the high back unrounded u; that merger began in the 15th century and for the central dialect of the
texts it was completed during the 16th century (Cf LKM 1972a:118). With the exception of a few
words such as hoik (hoi k o'lwo 1518 Sohak-cho 10:23b) > hulk ( hul'k ulwo 1586 Sohak 6:122a), in
word-initial syllables what o merged with was a but under certain circumstances in certain dialects o
was replaced by wo - in Phyengan (Kim Yengpay 1984:67) before a labial consonant or a syllable with
a rounded vowel so that nom > nom ‘other person’ (nam) and non'hwo- > nonwu- ‘divide’
(nanwu-). The standard language includes a few such cases: so'mayl so'moy > swomay (1617) >
somay ‘sleeve’. And there are also several words where o became e, such as ••• 'pol ‘time(s), layer(s)’
> ay pel ‘first (in time/order)’, pol'(s)sye I pol'sye > pelsse [dialect palsse] ‘already \po li- > peli-
‘discard’, to ll- > teli- [dialect] > teyli- ‘bring a person along’. (The regular development for the
verbs is found in the dialect versions pali- and tali-.) Writing the old vowel, called a lay o = alay a,
persisted in conservative spellings long after its distinctive value was lost. As ’Yu Huy (1824 Enmun-ci
12) observed, “o is confused with a (as A ‘child’, sa ‘fact’) or with u (as hulk ‘earth’)”. Symbols were
A Reference Grammar of Korean
PART I 43
created to write the syllables *yo and *yu, though the sounds did not exist in the language of the
writers of the early texts. It is surmised that those syllables had existed in a pre-Hankul version of the
language and were lost in the central dialect not long before the creation of Hankul in 1445. There is
evidence in dialect forms to reconstruct *yo for a number of words.
The created symbols are found in “1! kyo 11 kyu" (1446 Hwun 26a); later the letters
» yo and »>] yoy were created and exemplified with the syllables T7 kyo and kyoy
( ? 1750- Hwunmin cengum wunhay 16a), and still other symbols such as j4= yw(y)a and
rH yw(y)e were used to write Chinese sounds. CF Ledyard 1966:253, LKM 1972a: 126.
In the 15th century the vowels ey ay oy uy were articulated as diphthongs, and wey way woy way
were treated as triphthongs. For woy and way this statement may be questioned, since we assumed that
the Hankul symbols corresponding to the digraphs wu and wo of our Romanization represent simple
rounded vowels, unlike the diphthongs represented by wa and we. But I am now prepared to revise that
assumption and propose that the rounding represents a functional semivowel even in wu and wo.
Later the diphthongs got monophthongized, but the syllable uy itself, which had become u when
initial and i when not, was partially restored at the beginning of a word as the dissyllable ui by
younger educated speakers following the spelling tradition for the Chinese morphemes it represents.
The modern pronunciation of the genitive marker as /ey/ (the same as the locative marker) may reflect
a raised version of /ay/ from the MK allomorph 'oy rather than a lowered version of */i/ < 'uy itself.
The triphthong wuy became the diphthong wi, further monophthongized to a high front rounded vowel
[u] by some speakers. The triphthong way was reduced to the diphthong way [5e], and that was further
monophthongized by some speakers to the relatively rare articulation of a low front rounded vowel [5],
just as wey > wey was monophthongized by some to the mid front rounded vowel [6]. In modern
Seoul ay has merged with ey except when word-initial, where the distinction is maintained. In much of
the south the two are merged in all positions, and so are u and e. It is hard to say just when the
process of monophthongization took place in various parts of the country. The diphthongs can still be
heard as such in parts of Hwanghay, as noted in §2.2, and probably elsewhere, too.
The unrounded u was distinguished from wu even after labials (p ph pp m) until about 1748, but
then the vowel was assimilated to the labial, so that the modern mul (with “u” an abbreviation of
/wu/) represents the two distinct MK syllables mul ‘water’ and mwul ‘crowd’ (for which the modern
word is muli, with the accretion of the suffix -i).
It is generally assumed that earlier the modern back vowels u and e were articulated nearer (if not
all the way to) the front of the mouth. The Seoul pronunciation of u today, in fact, is fairly far forward
(central rather than back), and the sound of e has moved lower and toward the back, getting rounded to
something like [o], which approximates the sound we assume for the lost MK vowel o. Kim Wancin
has proposed that the MK vowels were quite different from the modern values (in all dialects) because
a Great Vowel Shift took place. He claims that the shift moved to the back of the mouth those vowels
that once were front and that it raised an earlier mid-central shwa *[o] to the value of the modern u,
while the former *[e] shifted back to the shwa position which is still heard for e in many dialects,
though Seoul has moved the vowel on toward the back low rounded version, apparently during the past
fifty years. Linguists in Korea generally accept the notion of the vowel shift, it seems, but do not agree
on its timing. Together with some of them, I favor retaining values for the vowels of the 15th century
that are close to the modern values, while I reserve judgment on the validity of the vowel shift at an
earlier period.
2.10.2. The earlier initials.
The language of the 15th century had all the initials of modern Korean with the exception of the
initial geminates pp tt kk cc ss. In addition the scribes wrote several kinds of clusters. There is general
agreement that initials spelled “ pC •••” began with a labial stop. And everyone assumes that there was
an oral obstruent at the beginning of the odd word sna 'hoy (1447 Sek 19:14b) ‘man (male)’, probably
contracted from sona'hoy (not attested until perhaps 1517), the source of modern sanay, though some
(such as LCT 121a) would interpret this as [tn] rather than [sn]. Yet many scholars (including LKM
44 PARTI
A Reference Grammar of Korean
1963:19) have doubted the face value of the initial sibilant written in the clusters “sp- st- sk-” and
claim that these strings had a pronunciation identical to that of later “pp tt kk”, the tense and crisply
unaspirated stops we refer to as “reinforced”. The geminate spellings of these initials were used in the
earliest texts only to present somewhat artificial readings for Chinese characters, in an attempt to
capture Middle Chinese distinctions that were ignored in nativized borrowings. They were sometimes
used also for phrase-internal strings that represented the prospective modifier -( u fo)lq -l- a voiceless
obstruent initial in the following word (especially when it was not a free noun), though these strings
were also given other treatments: (1) they were simply ignored, as often in 1481 Twusi and in modern
Korean; or (2) they were written with a cluster of final l + q, a symbol which otherwise wrote the
glottal-stop initial of traditional Chinese readings. And the digraphs ss and hh were used not only to
maintain traditional Chinese distinctions but also to write the initials of a few native words that had
perhaps incorporated an emphatic prefix. The only such examples of hh- are the verb stem (h)hye-
‘pull; ... ’ and compounds that incorporate it. I believe that by and large the textual spellings of native
Korean words must be taken at face value, and that when the scribes wrote geminates, including the
ss- and hh-, they were pronouncing them tense and with the same crisp release into the vowel that is
heard today. When they wrote sC- (or -s C- or -s-) they heard a sibilant articulation. The initial
clusters include not only the groups pt- pth- ps- pc- (no *pk-) and sp- st- sk- but also pst- psk-.
Those reluctant to allow the sibilant clusters treat these as equivalent to pt- and the missing *pk-. But
that makes it hard to explain the spelling contrasts found in 'ptoy ‘dirt’, pstay ‘time’, and sta[h], to
say nothing of the many strings that attach the genitive particle s in a phrase N s N to either the
preceding or the following noun, or run the words together with no junctures, as indicated by variant
spellings (see Martin 1982/3 and ? 1991). LKM, however, rejects the notion that reinforcement in
obstruent clusters was automatic in the 15th century (as it is today) and upholds the view that the
orthographic pt- contained a lax apical stop in contrast with the tense version in pst-. I believe that
the “sC-” clusters were pronounced with a sibilant both when initial (as syllable onsets) and when
medial (as interludes). The obstruent after the sibilant was unaspirated (as in modern English) and
identical with the reinforced obstruents of modern Korean, so that we can think of them as “spp stt
side”. The tenseness of the obstruent is a feature of the clustering of two obstruents, of which the
minimal case is two identical obstruents, the geminate pp tt kk cc ss, as indicated by spellings such as
-l 'tta for -Iq 'ta ‘will you ?’. What happened later (in the 16th century) was simply a suppression
of the s that left the tense allophones of the simple stops newly standing in contrast with the lax
allophones that were now the only version of p t k and also c and s. The change was probably gradual
and took place as the corresponding interludes became -tC(C)- in accordance with the merger of
syllable-final -s and •••/ which slowly took hold in the course of the 16th century, after a few earlier
harbingers. There was occasional dropping of p from initial clusters in a few attestations of the 15th
century; by the middle of the 16th century that was more prevalent. The dropping of the sibilant in the
sC (= sCC) clusters was probably complete by 1632, but it is unlikely that it started until the internal
-sC- had become -tC- (= -tCC-) in the middle of the 16th century; it probably got under way around
1575. Not all of the modern geminate consonants go back to clusters as such; some were created later
as emphatic versions of words, such as the sporadic appearance of forms noticed by Kim-Renaud
1977:92 in casual speech accompanying an emotional connotation. There were a few early verb
doublets with (s)C- that are thought to reflect a similar connotation added by tensing the initial (id. :93,
Ramsey 1978b:64), and the emphatic version is preserved as the modern stem: kku(s)- < skuzu- (1463
Pep 7:91a) < kuzG- (1463 Pep 2:200b) ‘puli’, ccih- < stih- (1466 Kup 2:62b) < tih- (1459 Wei
17:19a) ‘pound’, ssip- < ssip- (1462 'Nung 5:46a) < sip- (1462 'Nung 8:138a) ‘chew’, ssu- < 'ssu-
(1447 Sek se:4b) < su- (1465 Wen 2:2:2:41a) ‘write’. The doublets have been used to buttress the
argument that the sC- clusters did not contain a sibilant, at least in these words, but it is quite possible
that an emphatic prefix may have had the sibilant pronunciation.
With respect to the readings of the Chinese characters, LKM 1963:20 says that up until 1480 the
orthography used experimental elements, but then gave them up and simplified the spellings to be more
A Reference Grammar of Korean
PART I 45
natural. He finds that the natural readings began with 1496 Cin.en kwenkong, since 1481 Twusi gave
no readings for the many Chinese characters it uses. Perhaps that is why it is said that orthographic w
and Q were last seen in 1467. But that date is a bit too early, for texts from 1475-85 contain examples:
•••w • KYWOW-SYWUW (1475 Nay 2:2:69b) ‘teach’, "ttwow (1482 Kum-sam 2:3a) ‘way’,
PPEN-"nwow (1482 Nam 2:6ab) ‘agony’, ~PPWU-~MWUW (1485 Kwan 7b) ‘parents’
Q- ■QTLQ-'POYK (1475 Nay 2:2:72a, 73b) ‘100’, 'QJLQ-SOYNG (1482 Kum-sam 2:1b)
‘life’, 'QAK (1482 Nam 1:77a) ‘evil’, LI-'qyek (1485 Kwan la) ‘gain’
-LQ CIN-'SSHQ (1475 Nay 1:47b) ‘true’, ' QtLQ-'TTl (1482 Kum-sam 5:18a) ‘one
ground’, 'KAK-'PPYELQ (1482 Nam 2:63b) ‘particular’, 'hwolq-zyen (1485 Kwan
9a) ‘suddenly’
And it should be noted that 1481 Twusi wrote -lq C- in at least one passage: - se nilGwuylq ta
(23:44a) ‘will you be able to achieve the writing of ... ?’. But usually the reinforcement after the
prospective modifier was simply ignored by this work and later texts, with occasional exceptions: "kin
[KWA-'KUK] 'ul "mat kkwo (1481 Twusi 10:27b) ‘will they give up the long spears?’; "hwol tt i.n i
’’la (1588 Mayng 13:1b) ‘will do’. There are examples of -lq C in virtually all of the texts from 1445
to 1462; and also ? 1468" Mong, with -lq t, -lq c (15b, 24a), -lq s, -lq st, and even -lq h (5b); and
one exceptional case in 1475 Nay (puthulq 'ka 1:2a). But 1463 Pep, 1463/4 Yeng, 1464 Kumkang, and
1482 Kum-sam have only ••■/ CC for *-lq C, and the reinforcement is totally ignored by 1465 Wen,
1466 Kup, 1482 Nam, and 1485 Kwan. The texts that use -lq C also have examples of -l CC to
varying degrees, but the admirably morphophonemic spelling of 1449 Kok has only -l ss oy and -l
ss ye (it lacks an example of *-lq s), and uses -lq C for all other cases. The glottal was also used by
one or more of the texts in -lq st, -lq sk, -lq psk, -lq pst, -lq pt, -lq th, and -lq kh. Since in a few
instances it was used even before a voiced initial such as n or wo, the second symbol of the -lq string
perhaps sometimes reflected nothing about the pronunciation, but just helped identify the ending.
A number of Hankul-written words began with initial /•••. Some, such as ", Iwongtam ‘joke’ and
Iwo'say ‘mule’, are known to be of Chinese origin; and others, such as la'kwuy ‘donkey’ and le'ngwul
‘raccoon-dog’ are probably borrowings from Mongolian and Tungusic languages. The adnoun la'won
(1463 Pep 5:202b) ‘joyful’ - 'lawon in 1481 Twusi 7:25 - may be a contraction of 'LAK 'hwon ‘ ••• that
is joyous’. The /••• in these nativized borrowings and the many Chinese words beginning with L- were
probably not distinguished from n- by most Koreans of the 15th century, just as the two initials were
not kept apart later: the Chinese loanword loy-'ZILQ (1459 Wei 7:16a) ‘(to)morrow’ was assimilated as
noy'zil (1482 Nam 1:40b). But there may have been speakers who kept initial /-• distinct from n-:
1898 Tayshin has [rasil] for ‘tomorrow’.
An initial z- occurs in a few nativized borrowings from Chinese, such as zywoh ‘mattress’
(Middle Chinese nhywok), as well as in readings of Chinese characters, always as ZY- or zi-, and a
couple of mimetics ( zel-zel, zem-zem). The origin of 'zywuch > yuch ‘the four-stick game’ is
unknown, but I suspect that the shape was earlier *nywusk or *niwusk, in view of the Hamkyeng
dialect versions (nwus, nyukku, nyukkwu, nyukki, yukku, yukkwu, yukki, yuchi); a variant of
ne(yh) ‘four’ may be the first part of the word. It is unclear whether, or how, initial z- (or Z-) was
pronounced, but it continued into the 16th century before eventually disappearing. Internal -z-
resulted from lenition of s (including s from *c), often at the end or the beginning of a morpheme or
forming a morpheme in itself. A few nouns seem to have had an intrinsic -z- that remains to be
explained, though I believe that they too contain lenitions from s or, in the following words, *c:
mozom > maum ‘heart, mind’, kye’zulhlkye’zul > kyewul ‘winter’, kozolhlkozol > kaul ‘autumn’,
... . (The second syllables of these words were written with affricate-initial phonograms in 1400+
Kwan-yek.) As Ramsey has shown, the -z- verb stems come from underlying ~z%- (see §8.2.5), and
that in turn I believe is the result of leniting an s, under conditions as yet unclear. See §2.11.4.
In modern Korean a phonetic syllable can begin with the velar nasal provided it is between
vowels, but with a few marginal exceptions the -ng is final in a morpheme or in a syllable of a
polysyllabic morpheme. There are a couple of MK morphemes with the initial ng-: the polite marker
46 PART I
A Reference Grammar of Korean
-ngi and a bound noun ngek ‘place’, as in kunge'kuy (= *ku ngek uy ) ‘there’. But words and phrases
did not begin with the velar nasal. The traditional initial NG- was written for Chinese readings, but
when the words were nativized the initial was omitted, as was the glottal initial Q-. There are at least
25 examples of NG- that are illegitimate from the standpoint of seventh-century Chinese or of
Sino-Japanese. Virtually all begin with ngw- or ngyw-, as in ngwang ‘king’, "ngwuw ‘exist, have’,
and "ngyweng ‘longlasting’. But a glide is uniquely lacking in the perfective particle "nguy (1451
Hwun-en 2a), and while Tung T’ung-Ho reconstructed a voiced velar fricative for the Old Chinese
initial of this particle, Karlgren had treated it as z. The Old Chinese reconstruction for all the ngw-
and W~ words, based on shared graphic components in etymological sets, assigns them a voiced
velar - a stop in Karlgren’s system, a fricative in Tung’s. The character meaning ‘do; serve’ has two
readings in the texts: NGWUY and 'WUY (sometimes "WUY), and the nativized version is "wiry (1446 Sek
6:7b, 24a). The colloquial pronunciation of the 15th century lacked the means to cope with the
prescribed distinction of ngyen ‘polish’ : YEN ‘extend’ : QYEN ‘smoke’, or of 'NGWUY ‘guard’ : WUY
‘position’ (also "WUY) : 'QWUY ‘entrust’ (also "QWUY), with the resulting homonymy found in today’s
yen and wi. Kono (1968:17) says that in Korea nga was pronounced like (Q)A, presumably as lal, from
very early, since there is a (mis)spelling with the phonograms nga-TWO for the name of the priest
called A-TWO who came to Korea from China in 375. Quite a few nativized fish names end in -nge and
that represents a retention of the initial of nge ‘fish’ when not at the beginning of a word, as in "linge
(1518 Sohak-cho 9:25a) = "linge (1466 Kup 1:52b) ‘carp’; chyenge ‘herring’ (1799 'Nap-yak 27b)
must go back to *chyeng-nge (unattested as such, but see 1527 Cahoy 1:11 a=20b), with the two
nasals simplified to one. In the 16th century the symbol for the velar nasal came to be written only at
the end of a syllable and fell together in graphic shape with the zero initial with which it was in
complementary distribution.
2.10.3. Palatalization and dispalatalization.
In the 15th century the phoneme c and its aspirated counterpart ch were affricates, as they are
today, but they were not palatalized. The realization of c was as [ts—] or (§2.10.6) [-dz-]. The
palatalized articulations of the apicals in syllables such tye thye cy chye sye nye lye was an anticipation
of the glide and must have been present also in ti thi ci chi si ni li, as contrasted with tuy thuy cuy chuy
suy nuy luy. Because of examples of hy- and hi- that turn up in various dialects as s(y)- and si we
can probably assume that hye and hi (perhaps also the uncommon hhye and hhi) were palatalized.
Later, all these articulations underwent divergent developments in different parts of the peninsula.
In the south the nonpalatalized affricates were palatalized: ce merged with eye and che with chye,
so that there are two modern syllables ce and che, spelled “eye” and “chye” only when they are
contracted from “cie” and “chie”. Then the palatalized stops were affricated: tye and ti merged with ce
and ci. (The modern Seoul ti is a monophthongization of tuy, a raising of tey in specific words, or the
result of borrowing foreign words; and tye is for the most part a shortening of ti(y)e < tuy(y)e.) The
syllable si was made or kept palatal (a single frontal articulation, so not to be described as
“palatallZED”); sye got dispalatalized and merged with se, but was reintroduced to represent the
shortening of si(y)e. At the beginning of a word, ny- and ni- (including ly- and li- pronounced as
ny- and ni-) dropped the apical articulation and merged with y- and i'~, and they are so written in the
standard orthography of the south. Modern Seoul word-initial ni is the result of monophthongizing
nuy, raising the ney of specific words, or foreign borrowing. The suspective ending in Phyengan is -ti
from MK - 'ti, though the influence of Seoul has made the palatalized version -ci quite popular (Kim
Yengpay 1984:100).
In much of the north the affricates stayed apical, with no frontal coarticulation except before y or
i. But in those cases there was an erosion of the palatal quality: eye merged with ce, and tye with te.
The dispalatalization extended to nye and lye, which were not differentiated from ne and le, with the
result that n yeca ‘woman’ is pronounced yeca by southern speakers but neca by many in the north,
where the spelling is standardized as “nyeca” instead of the southern “yeca”. Internally, Phyengan has
swulo for swulyo ‘completing a course’ and illwu for illu ‘topnotch’ (Kim Yengpay 1984:69).
A Reference Grammar of Korean
PART I 47
We are not sure just when these changes happened, though the affrication of ti and ty seems to
have taken hold around the turn of the 18th century (LKM 1972a:67-8). For the nasal, there are words
that have individual histories, and a few doublets existed already in the 15th century. The verb stem
n yeki- ‘deem’ was written nye'ki- in 1481 but earlier the spelling was ne'ki-, a form that persists in
1936 Roth 37, no doubt the result of his hearing of South Hamkyeng speech. The verb neh- ‘put in’
appears as early as 1466 (Kup 1:13a, 2:41b) but the prevalent version was nyeh- (as in 1447 Sek
9:21a), which led to modern yeh- in Kyengsang (and elsewhere); yet here Seoul uses the glideless
version neh- heard widely in the north. In the word for ‘yes’ Seoul also follows the northern form ney
rather than the southern yey. 1894 Gale writes (95) niaki ‘story’ (iyaki > yayki > yeyki) and (165)
has the passage “yeng ila howo it is called ‘nyeng’ (mat)” - note the initial “ny” in the gloss - which
must be a contraction of (n)ieng ‘thatching’, derived from °ni- > "ni- > ni- ‘thatch (a roof)’.
1902 Azbuka kept sy [S] distinct from s, as did 1900 Matveev; and 1898 Tayshin “tsiui” < MK
'cwuy ‘rat’ seems to intrude the glide y without palatalizing or deaffricating the c [ts]. (Azbuka has
haysye ?< hay fijsie for ‘did’.) 1894 Gale 65 gives -l syeng pwuluta and -/ syeyta ‘it is likely that’ (=
-ulq seng siph.ta).
Among early signs of palatalization: fronting of the vowel in a'cik (1463 Pep 1:14a, ...) < an'cok
(1447 Sek 6:11a, 1463 Pep 1:44a, ... ) ‘yet’ (? < *a'ni cek ‘not time’), ho.yem 'cik (1518 Mayngca
14:21b) = ho.yem cuk (id. 14:16a) ‘worthy of doing’, achim (1736 n Ye-sa 3:9; cited from LCT
522b) < a'chom (1447 Sek 6:3b) ‘morning’; the doublet wum'chi- (wum'chye 1462 'Nung 2:43b) =
wumchu- ( wum'che 1462 'Nung 2:40a) ‘huddle, shrink’. Notice also, without affrication, ti's ’i (1449
Kok 43) = ' to's 'i (1459 Wei 10:20b) ‘like’; kile'ki (1568 Sohak 2:49a) = ' kuy'lye'ki (1527 Cahoy
1:8b = 15a) = kulye'ki (1462 'Nung 8:121b) = kulye'kuy (1459 Wei 2:40b) ‘wild goose’. On the other
hand the front vowel of silh.ta is not attested until quite late; 1894 Gale (177) has sulkhwo ‘disliking’
< sulh- < sul 'ho-. Yet (99) he writes nucin for nuc.un ‘late’ and offers the option of hoi sti (= hal
tti) or hoi ci (= hal[q] ci) ‘whether to do’; in ~ul ci entyeng (to) (64) he has both affricated ci < ti
and unaffricated ty-, if the representations are taken at face value. But there are examples of t and th
in ? 1517- 'No that are affricated in the Kyucang-kak version of 1795 'No-cwung though not in the
Phyengyang kam.yeng version, which is older in its language: "tywohi ( ? 1517- 'No 2:66a) = tywohi
(1795 ‘No-cwung [P] 2:59b) = cywohi (id. [K] 2:61b) ‘nicely’. The same stem appears somewhat
earlier in cywoha 'yla (1763 Haytong 103); and the postmodifier thyey ‘pretense’ is written chyey as
early as 1730 (Chengkwu yengen 92). Examples of c written for ty and chy for thy are found in 1632
Twusi-cwung, according to An Pyenghuy 1957.
Palatalization of velars also took place, mostly in Kyengsang and Hamkyeng. King 19886:291
finds seven examples of velar palatalization [k] > [tS] in 1900 Matveev, including ciwo ‘long’ = kio
< kl-l-, cilumi ‘butter’ = kilum ‘oil, grease’, ciley ‘on the road’ = kil ey; also, with simple
affrication [k] > [ts] for the Id, cili = kil ‘road’. The word kimchi ‘pickled cabbage’ is a back
formation (by false analogy) from cimchi, widely heard in the south (and also in Hamkyeng), the
expected palatalization of tim'choy (1527 Cahoy 2:lla = 22a) from Chinese niM-"CHOY ‘soaked
vegetables’ (LCT 1971:46). A similar hypercorrection is responsible for the development of cemsim
‘lunch’ into the dialect variant kyemsim, which appears in 1894 Gale 164 with the gloss ‘dinner’. The
word goes back to f'Jtyemsim (1518 'Ye-yak 38a) and comes from Chinese "TYEM-SIM, which refers to
those Dim Sum tidbits that “dot your heart” at lunch time. A similar case: kyel (1898 Tayshin) for cel
< tyel ‘temple’ (King 1988b:295:nl8). Dialect chi corresponds to standard khi for three nouns:
‘winnow’ ('khi), ‘height’ ('khuy), and ‘rudder’, which has only the variant chi in the earlier
attestations (1527 Cahoy 2:12b=25b “mis also chi ” - of mis nothing more is found, perhaps < "miflj
s ‘pusher’). (1874 Putsillo also attests chi ‘rudder’.) So the standard version of khi for ‘rudder’ seems
to be yet another hypercorrection. Putsillo has kina- for cina- < "ti-°na- ‘pass by’.
For Cye(-) the palatal quality is often shifted from the consonant to the vowel, metathesizing the
glide so as to produce the mid front monophthong of the modern language: selmyeng > selmeyng,
Phyengyang > Pheyngyang, ... . (King 1990 has examples that argue for ye > yey > ey with loss of
the initial glide after the fronting took place, rather than metathesis of the glide.) In Phyengan (Kim
48 PART I
A Reference Grammar of Korean
Yengpay 1984:69-70) c(h)ye > c(h)ey with nonpalatal [ts-], and sye > sey; hy- dispalatalizes to h-
in general, as well as in hey- ‘ignite’ < (h)hye- (standard khye- = khi-). The word myech ‘how
many; a few’ is widely pronounced meych and that is the way we write it for the modern language in
this book. The dispalatalization of the syllables m y ey and ph y ey is recognized by the NK orthography
but not by the standard spelling in South Korea. Both maintain a distinction of kyey from key and
hyey from hey that is no longer part of the spoken language.
2.10.4. Nasal epenthesis.
A small number of words have forms with and without a nasal before an affricate. There being
nothing obvious about affrication and nasality that would motivate a sporadic insertion of that sort, we
wonder whether the form with the nasal is not, in fact, the basic form. But for verb stems such as
a(n)c- ‘sit’ and ye(n)c- ‘put on top’ several kinds of evidence led LKM 1964 to the conclusion that the
versions without the nasal are older.
Ramsey 1978a:54-6 gives a good description of the situation, and points out that for certain words
the nasal insertion happened only after a non-affricate had become an affricate. Thus te'ti- ‘throw’
picked up the nasal of modern Seoul tenci- only after the syllable ti become ci. And hwon'ca (1518
Sohak-cho 10:6a) ‘alone’ did not have a nasal so long as it remained howo'za < hoWo'za < *hopo(n)
za (see honca in Part II). Yet if the etymology is *ho[n] "pwun 'sa ‘just one only’, that already
contains the nasal - and has another that is elided; but if the etymology is *ho[n] po[l] 'sa ‘just one
layer’, the nasal is not expected. The adverb acik ‘(not) yet’ goes back to a'cik (1463 Pep 1:14a),
which is attested also as an'cik ( ? 1517- ’No 2:12a, ? 1517- Pak 1:64a) and an'cok (1447 Sek 6:11a,
1463 Pep 1:44a), a form surviving as the South Hamkyeng an'cuk cited by Ramsey, and it perhaps has
the etymology *a'ni cek ‘not time’. If the nasal were original in all (or most) cases there would be no
need to explain why it did not develop for more than a small number of the words with affricates. And
the variable elision of the nasal could perhaps be attributed to whatever motivates the liquid elision
before apicals, the MK suppression of stem-final l before t n c s. Somehow the nasal elision never
happened to mence < mwon'cyefy) (also mwon'coy) ‘ahead; earlier’ in most of Korea, but one dialect
in South Cenla has mocye (Choy HakJcun 139). There was a nasal in an earlier attestation, according
to the interpretation by Kim Wancin (1980:155) of the phonograms in hyangka 14:10.
The verb stem a(n)c- appears as az- or as- in the forms as.non (1447 Sek 19:6a) and az'nwo'la
(1462 'Nung 1:3b) but that represents the reduction of the syllable excess (-no -» -c- > -s-,
different from the modern -no -» -n-). If the stem had really ended in the lenited sibilant -z- <
- 'so- we would expect the rising accent on the first syllable of those forms, as in the similar forms of
‘seize’: "as.non (1481 Twusi 22:49b), "as.nwon (1481 Twusi 16:68b), "asno'n i (1459 Wei 7:46b [ni
miscarved as na\), "asno'n i ’’n i (1462 'Nung 9:40a). Similar remarks apply to ye(n)c-: the forms
without the nasal can be treated as reductions of -no, with no rising pitch for yes.no'n i ’ la ( ? 1517*
Pak 1:56a) and a surprising initial high for 'yes.non (1481 Twusi 22:36b). All modern dialects have
the nasal; Ceycwu alone is reported to have a doublet aci-lanci-. No modern dialect lacks the nasal for
enc- (nor does any show initial y™). Putsillo 1874:572 has three forms with the nasal ( ansswo ,
ansswukuy, ansswukey) and one without ( acuwo ). LKM observes that 1103 Kyeylim (#317) uses
phonograms interpreted as “ a-cek-ke-la (for anc.kela ‘sit down!’) and 1400+ Kwan-yek (#349) used
phonograms taken as “ a-ke-la ”, both without the nasal, which could have been noted with a
phonogram “an-”, used by Kwan-yek to write “ an-ta” (#389) for "anta (? = a'no'ta or = "al’ta)
‘knows’ or the one used by Kyeylim to write “an(-h ay kwopoy)” (#229) for an'h oy kowoy
‘undergarment’.
There is an additional mystery. For the well-attested stem mon'ci- ‘stroke’ there are examples of a
variant without the affricate: mo'nye (1459 Wei 21:133a), mo'nisi'kwo (id. 18:14a), ... . Modern
dialects all have the affricate.
NOTE: In the first entry of Ramsey 1978a:55 correct the Seoul form to enchi and the gloss to
‘saddle blanket’, corresponding to the earlier enchi ( ? 1720- Waye 2:17b) < e'chi (1481 Twusi 20:9b).
A Reference Grammar of Korean
PART I 49
2.10.5. The earlier finals.
Koreans of the fifteenth century had syllables that, like the modern syllables, could end in a vowel
or in one of the consonants p tk mn ngl but there were also syllables ending in an s that contrasted
with t and was surely pronounced as a sibilant. Among the “overstuffed” morphemes were nouns and
verb stems that ended in the affricate c, in the aspirates ch kh th ph and simple h, as well as clusters
such as Ik Ip Iph Im nc nh sk and a few others. The extrasyllabic element spilled over into the following
syllable when an ending or particle beginning with a vowel was attached; otherwise (before consonant
or juncture) it was reduced to one of the codas permitted to a syllable. In the case of simple h that
meant it was dropped in the “free” form, so that nwoh ‘rope’ was pronounced (and written) nwo unless
followed by a vowel-initial particle ( nwo'h o'lwo ‘with a rope’) or by the copula (nwo'h i'la ‘it is a
rope’); but when a particle or ending beginning with t or k was attached, the basic h emerged as heavy
aspiration so that nwoh + two was pronounced (and written) nwo'thwo ‘also/even the rope’, and nah-
+ -'kwo was na'khwo ‘giving birth’. Some texts, such as 1449 Kok, wrote the syllable excess
morphophonemically so that kwoc ‘flower’, for example, was always written the same, while other
texts wrote the phonemic form kwos when no vowel-initial particle was attached. But even the most
generous of the morphophonemic spellers wrote the phonemic forms of phrases with those morphemes
ending in a basic simple h and did so until quite modern times when (apparently around 1933) the
“h-pat.chim” was invented. But by then the /i-final nouns, dropping all traces of the h, had become
ordinary nouns ending in a vowel (so that no ‘rope’ behaved like no ‘oar’) and only the verb stems
required the h final. There is evidence for final -h, presumably so pronounced, in earlier phonograms
(1250 Hyangyak) for mah ‘yam’ and (1103 Kyeylim) for 'cah ‘foot(rule)’ and pafhj ‘straw rope’;
there is no later direct evidence of the h in pa[h], but King tells me the word is treated like other ~h
nouns that are exempted from the umlauting rule in Hamkyeng. Cf LKM 1972a:85-6.
The 15th-century distinction of -s from -t was lost during the middle of the 16th century (1576
'Yuhap 1:8 spelled “sis namwo" for the sit namwo ‘maple tree’ of 1527 Cahoy l:5b=10a), so that in
modern Korean the syllable-final phoneme -t neutralizes those two final consonants, as well as the
morphophonemic finals that were already neutralized in -t (-th and for some -Ith) and in -s (•- c ,
-ch). The affricates (~c and -ch) were distinguished in syllables written with phonograms in 1250
Hyang-kup (LKM 1972a:83-5), but they had fallen together before 1400+ Cosen-kwan, which wrote
words ending in -c -ch -s alike, with a Chinese character (THOUGHT) that must have represented a
sibilant. In transcribing Manchu and Mongol the syllable-final -s was treated as a sibilant (in contrast
with the stop -t) as late as 1748 (Martin ? 1991 :nl3).
The Hankul system of initials made provision to distinguish five kinds of “throat” sounds at the
beginning of a syllable: (1) simple vowel onset, using the zero initial; (2) sharp onset with the glottal
stop q-; (3) the nasal velar ng-; (4) breathy onset with h~; and (5) the reinforced (murmured)
breathiness of hh- . These were all needed to write the traditional distinctions of reading Chinese
characters, but speakers of Korean did not normally distinguish q- or ng- from the smooth onset;
1446 Hwun [25b] explicitly states that the glottal onset was not distinguished from the smooth in native
Korean words. And in the Chinese readings, the q- was traditionally distinctive only before i or a
glide: Ql(-) QY- qyw- qw- were supposed to be different from Y- YW- w- but for the other
vowels the q- was automatic so that the only versions were QA(-), QE(-), QO(-), QU(-). For the
Chinese readings only, the early spellings put a final zero (the circle symbol) below a syllable that
ended in a vowel, so that every Chinese syllable carried a pat.chim of some sort, but this practice fell
into disuse by the early 1500s. The zero (the circle symbol) was distinguished from the final -ng,
which had the teardrop shape with a tick at the top. When the open syllables of even the Chinese
readings came to be written like the open Korean syllables (with no pat.chim) and the differentiation
of initial ng- from zero fell into disuse (by the 1490s), the symbols for the zero initial and the final
velar nasal were placed in complementary distribution, and they ended up merged into a single symbol
with different realizations as onset (nothing) and as coda (velar nasal).
Among the -1C clusters, -Iq was written in non-Chinese expressions only for the prospective
50 PART I
A Reference Grammar of Korean
modifier ‘(~ that is) to do/be’, which had the effect of reinforcing the simple obstruents p t k c s (but
not h ) when they began a following noun in close juncture. The phenomenon was also written, as later,
by using the geminate clusters: •••/ CC- = -Iq C™. And after 1480 it was often left unwritten;
realizing that the liquid represents the final of the modifier, the scribe keeps constant the shape of the
noun and spells •••/ C-. So the coda -Iq represents a morphophonemic phenomenon rather than a string
of two phonemes. The -q anticipates the tense component of the reinforced obstruent, recognizing that
it is not part of the basic shape of the noun - nor, originally, of any other morpheme of the language,
since the modern geminate initials became phonemic only by grace of the initial p- and s- clusters. It
is likely, however, that the source of the morphophonemic peculiarity is an attachment to the modifier
of the genitive (= adnominal) particle s, as evidenced by the spelling -Is C- for a few examples of the
structure (see -ul s in Part II). The coda -LQ was also used regularly to represent the reading of
Middle Chinese unreleased final -t, the “entering tone” equivalent of -n in the other three tones.
There was apparently no difference in pronunciation from an ordinary •••/, as shown by some of the
words written as normal Hankul rather than character readings, but the creators of the system must
have realized there was something odd about this group of syllables. They may have been aware that in
this millennium some Chinese dialects, such as the language of Canton (Guangdong), retain the
7th-century -p »t -k while others (such as southern Mandarin) use a glottal stop for all syllables with
the entering tone, merging the three codas: -p -t -k > -q. (Still others, such as northern Mandarin,
lost the stops and merged the syllables into the other tones in various ways.) The creators of the system
of spelling out the readings of Chinese characters may have been baffled by the Korean choice of •••/
for the syllable coda when -t was available in Korean words; 1446 Hwun [22] explicitly notes that the
Chinese coda is properly -t but that it is popularly treated as •••/. A satisfactory explanation of that has
yet to be offered. Perhaps at the stage when the Chinese words were coming into the language Korean
lacked a syllable-final -t; the etyma with the MK final stop could have come about by shortening
forms with a final vowel (is the negative "mwot from *mwo 'to?), so that •••/ was the only available
coda similar to the Chinese, and then as now it was already a lenition of basic t in certain words.
Yet there are at least two facts that suggest the coda -LQ may not have been the same as /•••/ /. The
first fact: in forms of the verb ‘do’ the contraction of hok- to kh- (rather than ’k-) and of hot- to th-
(rather than to ?•••) takes place when the preceding phoneme was voiced (m n ng l w V Vy), but -LQ
was sometimes treated as if voiceless: ' kywelq 'key (not *khey) ho'n i ‘[he] let them decide it’ (1447
Sek 9:20a), just like na 'kot ’’key (1447 Sek 6:1b) and ko'tok ’’ti (1462 'Nung 1:67a). There are
counterexamples: thwong-'ttalq khe'tun (1463 Pep 1:9a) ‘if they are knowledgable’ rather than
'ke'tun (as in "mwot ’ke'turi), " kay-'thwalq 'khey (1459 Wei 21:48a) ‘so as to emancipate’, rather
than ’’ key . And *'ho-to'lwok > cywung-'ZILQ tho'lwok (1465 Wen se:5a); ho'ti > ' polq 'thi (1463
Pep 4:93b), 'TTAY-'ssiLQ thi (1464 Kumkang 38a), and even ' kywelq thi (1482 Nam 1:50a) contra the
earlier example. Other cases of / are treated as voiced: ecul 'khwo ‘is disturbed and’ (1447 Sek 6:3b),
'ecul 'khey 'ho.ya (1482 Kum-sam 2:19b) ‘making them disturbed’, ecul thi a'ni 'ho'ya is'nwon t
oy (1462 'Nung 1:69b) ‘when it is not disturbed’; ClN ol 'khwo eye ho rn ye n’ (1462 'Nung 7:73b)
‘when one wants to be true’. And compare modern kyel kho ‘absolutely (not)’ < ‘ KYWELQ + hfoj'kwo
‘deciding’. The second fact: in Chinese binoms the modern language always reinforces an initial t-,
c-, or s- of a morpheme which follows a morpheme that ends in -1 (see §1.5): palqtal = /palttal/
‘development’ < ‘ PELQ-'TTALq . Since this reinforcement has traditionally been ignored in the Hankul
spellings, we presume that it continues the articulatory habits of the earlier language.
The MK voiced fricative W was a phoneme that represented the lenited form of p (§2.11.2). As a
coda the symbol was also to write the labial glide (postvocalic -u or -o) that was traditional to certain
Chinese syllables, but in the pronunciation of Koreans the letter was simply ignored.
The lenition of k (§2.11.1) was shown by a device that we interpret as the phoneme G, which was
distinctive only after l z y i. (The device blocked the liaison that normally would make the l or z a
syllable onset and would insert a y before a vowel following y or some cases of i.) After the 1400s G
lost its phonetic effect, as shown by absence of the device in some of the spellings of 1527 Cahoy,
such as ke'zwuy (> kewi, dialect kesi) for earlier kez'Gwuy ‘intestinal worm’ (Cf LKM 1972a:86).
A Reference Grammar of Korean
PART I 51
2.10.6. Intersyllabic strings; assimilations; conflation and compression.
The interludes between syllabic nuclei (vowels) of modern Korean comprise all the possible
sequences of coda + onset, but with assimilatory adjustments that merge morphophonemically distinct
strings, so that ip man ‘just the mouth’, iph man ‘just the leaf, and im man ‘just the beloved’ are
pronounced alike as /imman/. Variant spellings sometimes indicate an understandable confusion about
the forms that make up certain words: ancumpangi (1894 Gale 183) = anc.um pangi is written for
anc.un payngi ‘cripple’ and pipin pap (1965 Dupont 137) for pipimq pap ‘rice hash’.
In MK spellings nearly all expected strings are found, so that there are quite a few interludes. To
these are to be added the many strings created by the genitive (= adnominal) marker s. That particle in
the structure N s N was handled in three ways: it was written as the coda (or part of the coda) of the
preceding N, it was instead attached to the following noun as the onset (or part of the onset), or it was
placed all by itself as a graphic syllable in its own right. The extreme case is probably tolks pstay
(1446 Hwun [25b]) = tolk s pstay ‘the Hour of the Cock’, a phrase that was probably pronounced
with one or two junctures, if all consonants were fully articulated. Comparable pronunciation problems
in English can be found in “the milk’s splatter”, “sports credits”, and “a barely glimpsed strain”.
Whenever possible a syllable of Korean begins with an onset, so that if a morpheme that ends in a
closed syllable is put before one that begins with a vowel, its coda becomes the onset of the second
syllable and the first syllable becomes open. That is why ip i ‘the mouth [as subject]’ sounds like i pi
‘this rain (or broom)’ when the latter is pronounced without a juncture. This liaison phenomenon
seems to have been present in the language as far back in time as we can go. When a spelling retains as
coda a movable string and uses the zero circle to write the onset of a vowel-initial second element, we
know that the string was written morphophonemically or includes a juncture - or (before y or i) that it
represents IGI (§2.11.1).
The assimilatory phenomena that affect the interludes quite often lead to the merger of
morphophonemic strings, as we have seen in the replacement of voiceless oral + nasal by the
corresponding nasal + nasal, but some are at an allophonic level - they are purely phonetic.
Voicelessness is a nondistinctive feature in Korean. Obstruents are voiceless after or before juncture or
when clustered with another obstruent (including h). Otherwise they are voiced, except that (at least in
modern Seoul speech) s and h are intrinsically voiceless and, both alone or clustered, spread the
voiceless stretch over most or all of the syllable - the vowel and often a final -m -n -ng - 1 , as well.
The contrary tendency is for the simple h to get murmured and dropped between intrinsically voiced
sounds (vowels, m n ng 1). On the possible voicing of s to [z] in dialects and older varieties of the
language, see §2.4, §2.11.4. The lenition phenomena of earlier Korean appear to support the notion
(Martin ? 1991 :n2) that the MK simple stops and the affricate c were voiced between vowels, as they
are today, and that the lenitions taking place were just a matter of the voiced stops weakening to
fricatives. Since s was already fricative, however, a distinction between the unlenited and the lenited
forms must mean it was voiceless when not lenited, assuming that the spelling indicates the situation
with reasonable accuracy. What about the allophones of 11 It is quite possible that the flap version [r]
was common as a coda and that the lateral developed as part of the tendency to foreclose the release of
consonants in the coda when they could not be moved over to become the onset of a following syllable,
but the timing of both matters is uncertain. In any event, recognition that lateral articulations exist is
made in 1446 Hwun ([25b]) which says “light and heavy l are not distinguished in Korean” and “there
is only one initial in Chinese syllables”, but suggests making a letter for the light liquid (kapyewun
liul) by putting a circle below l (like that put below p or m to write W). The symbol was adapted by
Korean scholars writing Manchu and Mongol glossaries of the 1700s, who put the little circle to the
right of the syllable with the /; the circle was used also to mark other peculiar phonemes, so that it
functioned as a kind of asterisk or pointer rather than a mark specific to the liquid.
52 PART I
A Reference Grammar of Korean
Evidence of the nasal assimilation rules became more common in spellings (or misspellings) of the
late 1600s, but there are examples I have seen of -pn- -* -mn- from 1586, of -kn- -* -ngn- from 1553,
and of -tn- -*• -nn- from as early as 1481 and perhaps one from an unavailable text of 1466 (Mole 1;
cited from LCT 406b): punnon *- puth.non ‘igniting’. Noun stems ending in -h dropped the coda
before a nasal (as before a vowel), so that wuh + "ma'ta -* wu "ma'ta (1447 Sek 6:31a) ‘atop every
one’, but that may be because all nasal-initial particles were loosely attached. With verb stems,
however, the processive -no- attached tightly, and a stem-final -h- was realized as either t or n :
il'hwum cit.no'n i ’’la (1459 Wei 2:49b) ‘affixes a name’ ■*- cih-, cet'nwon t ol (1462 'Nung 2:54a)
‘(the fact) that one fears it’ *- ceh -; cwonno'n i (1463/4 Yeng 1:59b) ‘is quite pure’ *- cwoh-,
nwot'nwo'n i (1447 Sek 13:19ab) ‘sends one off’ *- nwoh-. Some early texts used the unique
syllable-initial geminate nn- to write the result of -h- + -n-, as in 'han "swum ti.nnon swo'li (1447
Sek 19:14b) ‘the sound of uttering a deep sigh’ and ta.nno'n i ’’la (1451 Hwun-en 15a) ‘(it is that) it
touches’ tah-. When the stem ended in -Ih- the result was the unique string ••• l.nn -, as in il.nno'n i
’’la (1462 'Nung 2:2a) ‘(it is that one) loses’ *- ilh-. The same text will also write il.no'n i (1:62b)
‘(that one) loses’, so perhaps the difficult pronunciation iln-no- was simplified to just il-no-, since
there was no other need for syllable-final -In. There are also rare spellings of -Ih.n- as in halh.no'n i
(1462 'Nung 8:5b) ‘(it is that one) licks’, retaining the basic morphophonemic shape. But the most
common spelling is -l.n-, as in tung ul al'nwo'n i (1459 Wei 2:9a) *- alh- ‘has an aching back’ and
sul'nwon 'pa ’yl's oy (1481 Twusi 8:7b) *- sulh- ‘because it is distressing’.
In the modern standard language the liquid dominates the nasal in the morphophonemic strings
-In- and -nl-, both of which merge with •••//•••, pronounced as a long lateral with lateral release.
Misspellings indicate that this merger began in the 1700s. In the 1800s there was a tendency to write
“ l.n ” for IWI regardless of the etymology.
Variant spellings indicate that some of the different consonant strings written between vowels were
phonetically equivalent. The difference in spelling is due to (1) considerations of the basic shape of the
morphemes juxtaposed, (2) attention to compression or conflation under differing speeds of articulation
(tempos), (3) misinterpretations and indecisions stemming from other factors. In the case of simple
interludes, the placement of the syllable boundary (as indicated by the dots in the Romanization) was
irrelevant to the pronunciation except when it indicated a juncture, usually shown by space or hyphen
in the Romanization. The following sets of orthographic strings were phonetically equivalent in Middle
Korean (a dot shows the syllable boundary):
s.p
= .sp
p.t
= -Pt
ns.k
= n.sk
Ip.s
= l.ps
s.k
= .sk
p.th
= .pth
ns.t
= n.st
lp.sk
= l.psk
s.t
= .St
p.s
= .ps
ms.k
= m.sk
Ip.psk
= l.psk
s.s
= .ss
p.h
= .ph
ms.t
= m.st
Ip.c
= l.pc
s.G
= .zG 1
k.h
= .kh
m.psk
- m.sk
Ip.t
= l.pt
t.h
= .th
m.pst
- m.st
Ip.h
= l.ph
p.sk
= .psk
Ik.k
= l.kk = Iq.k
.nn = n.n
l.nn = l.n
1 As in pos'Ga (1447 Sek 6:31a) = poz'Ga (1462 'Nung 1:5a) ‘crush’.
Through the centuries spellers have been plagued by the fact that when the tempo of speech slows,
the reinforced and aspirated consonants are anticipated by closing a preceding open syllable: appa and
apha become ap-ppa and ap-pha, kacca and kacha become kat-cca and kat-cha. (But you will rarely
hear the word /isse/ slowed to become *it-sse or /mosse/ said as *mot-sse.) Since each of these strings
can represent several morphophonemic strings, the speller has to pay attention to the morphemes to
know whether to write t.h, t.th, .th, th., or h.t - or even th.h, h.th, th.th (though the morpheme
structure makes those particular spellings unlikely). In rapid speech these strings will all be
compressed to just /th/ and in slower speech they will all be conflated to /tth/. Texts of the past
several hundred years have many examples of misleading conflated forms such as that of 1894 Gale
A Reference Grammar of Korean
PART I 53
cip.phoyngi for ciphayngi ‘staff’, and those cited in Martin 1982, to which can be added an earlier
example of perhaps a similar sort (assuming a scribal interpretation of nyek as nyekk): twong nyek kuy
[= nye'kuy] chi l’ (1466 Kup 1:21b) ‘the ones from the east’; compare twong nyek 'kwo'lwo [= nye k
wo'lwo] (1518 Sohak-cho 9:98a) = twong 'nyek khu'lwo [= 'nye'kh u'lwo] (1586 Sohak 6:91a) ‘to
the east’. Earlier texts also have examples of misguided morpheme divisions such as 1894 Gale pip.ye
(162) = pipye ‘mix’ and nyek.yes.ci.wo (113) = n yekyess.ci yo ‘deemed’, as well as etymologically
motivated examples such as cip.wung (1936 Roth 42) ‘roof’ and pak.aci (1881 Ridel 166) ‘gourd
dipper’ (so spelled also in 1632 Kalyey 4:20a).
When the reinforced consonants emerged as phonemes in their own right, the i which had so often
preceded them was widely used to write sp st sc sk where the modern language has chosen to
institutionalize a different device, less common earlier, the geminate pp tt cc kk. Since the syllable-
final -s had merged with -t in the 16th century, a majority of the morphemes that are heard in certain
environments as -t are written ~s because in other environments (before a vowel-initial particle,
ending, or copula) they have the sibilant pronunciation. These morphemes, too, get conflated in slow
speech as shown by such spellings as wos.si (1894 Gale 99) = os i ‘clothes [as subject]’ and stus.sun
id. 109) = ttus un ‘as for the meaning’, in both of which the interlude represents a long sibilant. In
nas.cun (id. 110) = nac.un ‘low’ and in is.hun nal (id. 115) = ithun nal *- ithut nal ‘the next day’
the s represents an apical stop: /natcun/, /itthunnal/. In the case of pas.sol (id. 104) = path ul ‘the
field [as object]’ the slow form /passul/ tells us that Gale was hearing a dialect that had simplified
some of the overstuffed nouns, path -* pas.
When foreign words have an interlude spelled “-tt-” Koreans like to use the Hankul final -s to
represent the first initial: Los.ttey ‘Lotte’, Cheyusu Maynhays.then Unhayng ‘Chase Manhattan
Bank’. This violates the unstated rule that the conflated form is to be written only if a morpheme
boundary is recognized. A corollary to the rule is that /ptt/ (or the like) is to be spelled p.t unless the
second morpheme begins with tt, but in the case of ipttay ‘up to now’ and cepttay ‘not long ago’ the
decision ignores the etymology (< i + 'pstay, tye + pstay), as it does in copssal ‘millet grain’ (<
cwo[hJ psol ) and similar words, and associates the second syllables directly with modern nouns ttay
‘time’ and ssal ‘grain’.
2.11. Lenitions and elisions; sources of G.
The fifteenth-century language of the early Hankul texts offers many examples of the weakening
or total loss of certain consonants between vowels. Some of the effects can be seen at the end of verb
stems or at the beginning of suffixes and particles. Others are internal to words or morphemes. The
Hankul system made provisions to write voiced fricatives for labial, velar, and sibilant categories.
These functioned as distinct phonemes W, G, and z in the language of early texts, but we have reason
to believe that other varieties of Korean of that day retained the p, k, and i which had been lenited to
create the passing distinction of the voiced fricatives. Many examples of these MK sounds turn up
unlenited in modern dialects, especially those of Kyengsang and Hamkyeng (Cf Ramsey 1975, Martin
1982/3, Kim Yengpay 1984:168-72): melkwu ‘mulberry, wild grapes’ (melwu) < melGwuy, molkay
‘sand’ (molay) < mwol'Gay ; masul ‘village’ (maul) < mozolh\ saypi ‘shrimp’ (saywu) < sa'Wi;
... . In the case of verb stems, the lasting effects of the lenitions can be seen in the shape alternations of
the -w- stems (— w-/—p-), the -(S)- stems (~s -/••• -), and some of the -ll- stems (— lu-/ — 11-). The -t/l-
stems, earlier as today, showed an alternation of the stop -t- before a consonant and the flap •••!-
before a vowel; the flap represents a lenition of the stop. Since certain dialects today do not have the
lenited forms for -W-, -(S)-, and -t/l- stems, we assume they are preserving paradigms of unwritten
forms of the fifteenth-century language that were closer to the original system. (An alternative
argument would say that these dialects have restructured the paradigms by analogy.) We believe that
the motivating factor for the lenitions was largely accentual, but the detailed circumstances remain to
be adequately described. Some of the patterns probably result from accompanying vowel elisions and
other factors inducing compression. The susceptible particles and endings mostly have velar initials
(-k- -* -G-), but notice also the bound stem -"zoW- (deferential), source of the modern -sup- that
54 PART I
A Reference Grammar of Korean
marks the formal style, and the particle 'za ‘precisely, only (if)’, still said (i)sa as a dialect version of
Seoul (i)ya, which comes from attaching that particle to the nominative marker 7, then eliding all
cases of -z- so that 'i'za became ia, which inserted the glide heard in iya. Elsewhere it is proposed that
the copula forms ila and (?)iley are lenited versions of i-ta and i-tey.
The phoneme G as assumed here neutralized the several kinds of lenition. The 15th-century
spelling distinguished the phoneme only after y, i, l, and z, but it seems likely that earlier it may have
been present between two vowels, especially when one or the other was the minimal (and often
epenthetic) u or o, and especially between y and u or o, where we will write yGu and yGo even though
there is no contrasting *lyyul or *lyyol: keyGulu- / keyGulG- ‘be lazy’, "nwuyGus.pu- ‘be remorseful’,
... . For a number of such words the source of the G can be found in dialect versions that preserve the
original consonant: keyGulu-1 keyGulG- ‘be lazy’ is keykulu- in North Kyengsang (Kim Hyengkyu
1974:368) and Hamkyeng (Kim Thaykyun 1986:55), and nxvuuy < *nwu[G]uy ‘sister’ not only has
dialect versions nwupay, nwupi, nwupu but was written with the phonograms “ nwu-pi ” in 1400+
Cosen-kwan. For a verb stem like towoy- ‘become’, where a missing consonant is suspected, we are
tempted to write, for example, “ to[G]woy -” with the understanding that the source of the [G] may be a
velar (a lenited k) or a labial (a lenited p), and occasionally even a sibilant (z = lenited s). In the case
of ‘become’ we know the missing consonant was labial because of the attested variant toWoy-, so we
can presume a history of *topoy- > toWoy- > *toWwoy- > *to[G]woy- > t[o]woy- = modern toy-.
But we will forgo writing [G] in these cases and use that notation only for the elision of the velar
initial of certain bound elements ( kwa, kwos, kwom; 'ka, kwo ; 'key, 'kuy; ... ) which appear with
the lenited velar (' Gwa , Gwos, Gwom\ Ga, Gwo\ - Gey, - Guy) after l, y, and often i. Such
notations as [G]wa and - [G]wo are offered as helpful reminders of the immediate sources of forms
with elision. But we will write ‘become’ as towoy-, and similarly leave implicit the likely dropped
consonant in these words, among others: e'[ Jwul'Gwu- ‘join them’, kuf Jwul- ‘act’, mwu[]u- ‘shake’,
no[ Jwoy('ya) ‘again’, sa [ Jol/sa hoi ‘three days’, te[ Ju- ‘increase’, ta[]o- ‘get exhausted’.
Because of the neutralization represented by the phoneme, Kim Cin.wu and To Swuhuy (1980)
treat our -G- as a juncture phenomenon rather than a segment holding specific phonetic content. And G
may very well have become a purely graphic convention in later stages of the orthography, before it
vanished altogether.
In our Romanization of modern Korean we sometimes indicate an elision with an apostrophe,
though not in paradigmatic forms. Among the elisions represented by the apostrophe are these: [k] in
'yu’-wel ‘June’, [p] in si’-wel ‘October’, [m] in camca’ kho ‘quietly’, [ng] in su’ nim ‘monk’, [I] in
cha’-cita ‘is sticky’, [i] in kac’-kac(i) = kaci-kaci ‘all kinds’, [e] in hal-’meni, [wu] in mak’-kelli
‘coarse liquor’.And sometimes the elision is of a syllable: [ci] in ape’ nim ‘father’, [ni] in eme’
nim. (The elision in these two expressions is ahistorical, for the etyma are ap% and em%.)
2.11.1. Velar lenition and elision.
Under certain circumstances the MK velar stop k lenited to the sound that we transcribe as G,
which was probably articulated as a voiced fricative (velar or laryngeal) or at least a glottal squeeze.
The sound was recognized only after y, i, l, and z (zG was often written sG). Hankul used indirect
devices to show it, blocking the usual liaison that would (1) make -l or -z the onset of the following
syllable and (2) accrete a syllable-initial y before a vowel after -y or -i.
The circumstances calling for velar lenition (Cf LCT 1961) involve the joining of noun + particle
or verb + ending. The endings include the gerund -'kwo, the adverbative -'key or its variant -7fc%y,
and those complex endings that incorporate the gerund or are built on the effective formative - ke-\ but
there are no examples of lenition (to *- Gi) of the summative - ki, which was little used at the time.
The lenition took place after all stems ending in -y-, •••/-, or -z- but not after most of those that ended
in •••;-. For an •••/- stem to qualify, it had to be:
(1) the copula i- , which predicates nouns. Examples can be found in these entries of Part II:
'i'Gen ma'lon, "iGe'na, 'iGe'nol = 'iGe'nul, 'iGen 'tyeng, 'iGe'nywo, 'iGe'tun, i'Gey, i'Ge 'za,
'i'Gwo, and their shortenings to 'y- and ’•••.
A Reference Grammar of Korean
PART I 55
(2) the causative "ti- ‘drop it’ (< °ti- ‘fall’).
(3) one of a few polysyllabic steins ending in - li- (such as spu'li- ‘sprinkle’, e'li- ‘be stupid’, ... )
that were probably confused with the structure -u'l i (copula prospective modifier 4- postmodifier
4- copula). But most of the -li- stems do not trigger the lenition: po'li'kwo ‘discarding it’, ki'li'kwo
‘praising’, ... . And the “confused” stems do not always lenite: e'li'Gwo (1462 'Nung 7:67a, 1463 Pep
2:242a) but also e'li'kwo ( ? 1517‘ Pak 1:9a) ‘being stupid’; no li'Gesi'nol (1445 ^ong 8; for
*no'li'kesi'nol = *no'lisike'nol ) ‘[the emperor’s command] came down, and ... ’, yet no'li[]kwo
(1481 Twusi 10:35b) ‘coming down’.
The peculiar behavior of the stems i- and "ti- led LKM to the conclusion that the basic forms
they represent are iy- and "tiy -; contrast ti'kwo (1445 ’Yong 86) ‘falling’ with "ti'Gwo = "tiy'Gwo
(1459 Wei 10:24b) ‘dropping it’. Independent motivation for that conclusion can be seen in the MK
abbreviation of the copula as ’y- after a vowel (where modern Korean usually suppresses i- leaving no
trace) and in the derivation of the stem "ti- from *ti-'[G]i- (intransitive verb 4- causative), as
confirmed by the accentuation.
Nouns and adverbs that end in -i generally triggered the lenition: "ne y 'i'cey swo'li Ga a'ni
Ga (1462 ’Nung 4:126b) ‘is it now your sound or isn’t it?’; i 'Gwa (1451 Hwun-en lb) ‘with/and
this’; ku li 'Gwos (1459 Wei 8:62b) ‘just/precisely that way’; 'wuli Gwos kyeyGwu'm ye n’ (1459
Wei 2:72a) ‘if we are the ones defeated’. But after the negative precopular noun a'ni ‘not’ and the
expression hon ka'ci ‘one kind = the same’ the lenition seems to have been optional:
tye non hwo'za "salom a'ni 'ka (1475 Nay 2:1:16a) ‘isn’t he a person alone?’; i ’ sywelq-'pep
ka, i ’ sywelq-'pep a'ni Ga (1482 Kum-sam 4:37b) ‘is this preaching the law or isn’t this
preaching the law?’
hon ka'ci 'ka talo n i 'ye (1459 Wei 8:31b) ‘are they the same or different?’; hon ka'ci 'Ga
a'ni Ga (1462 *Nung 1:99a) ‘is it the same or not?’
The obligatorily leniting nouns include most prominently the postmodifier i ‘the one that ••• ; the
fact that - ’ in all of its uses. Most examples involve the zero abbreviation of the copula stem, as seen
in these entries of Part II: -u'l i ’Ge m ye, -u'l i ’Ge n i ('Gwa, ’’ston), -u'l i ’Gen ma'lon, -u'l i
’Ge'nul, -u'l i ’Ge'ta, -u'l i ’Gwan'toy. I have been unable to find structures with *-u'n i ’G-. And all
of the examples of l-(u)'li'Gwol or /-(u)'ni'Gwol are questions (-u'l i 'Gwo, -u'n i Gwo, see below),
for there seem to be no such expressions with the copula gerund (*-u'l i ’'Gwo or *-u'n i ’’Gwo). In
structures with the postmodifiers 'ka and 'kwo ‘question’, the copula modifier ('in) is usually
suppressed: (-u'n i 'Ga, -u'n i 'Gwo; -u'l i 'Ga, -u'l i Gwo; -ke'n i Gwo) but there are examples
that let it surface, as found in the entries a'ni ’n 'ka (= a'ni Ga = a'ni ka), a'ni ’n 'kwo (= a'ni
Gwo = *a 'ni 'kwo), and 'in 'kwo of Part II. Also included: the suspective - ti (< t i, postmodifier +
nominative particle), as found in nwop'ti 'Gwos (1459 Wei 1:37b) ‘the higher they are ... ’.
The particles 'kwa ‘with’, 'kwos ‘precisely’, and 'kwom ‘each’ regularly lenite to G- after a
noun that ends in / (including lq), y, or i. The velar initial is totally suppressed after other vowels
(including vowel 4- w); we note this by writing the G in brackets, as a reminder to help identify the
morphemes. The postmodifier 'ka ‘question’, is regularly preceded by a modifier but the copula
modifier 'in is usually omitted, so that 'ka stands right after the noun, as if a particle, and it lenites
just like the other particles. The few exceptions written without lenition may be due to scribal error or
later restructuring:
na'la/hj s 'kul i "ta "hwo 'uy 'hoy'Gwon 'pa ka (1586 Sohak 6:41b) is a rewrite of na 'lafh] s
kul'Gwel i "ta chwoy-' HWo 'uy hon kes 'ka (1518 Sohak-cho 9:45b) ‘is the writing of the nation’s
history all by Cui Hao?’;
ha'nol kwa sta.h i khu 'kwo (1481 Twusi 21:2a) ‘heaven and earth are large’ should be ha'nol
Gwa (1462 *Nung 8:131b) or ha'nol'khwa = ha'nolh 'kwa (1462 *Nung 2:20b).
The example na kwos ( ? 1800 Hancwung 90; cited from LCT) ‘precisely I’ is from a late text. The
word cey'kwom (1518 Sohak-cho 8:3a) is a shortening of ceyye 'kwom ‘individually’, a derivation yet to
be explained. The adverb-intensifying suffix - 'kwom (which may or may not be the same etymon as the
56 PART I
A Reference Grammar of Korean
particle) never lenites: ta'si-'kwom (1447 Sek 6:6a) ‘again’, kwop'koy-'kwom (1459 Wei 1:47b)
‘double; twice (the age)’. Perhaps juncture could account for a'lwo'm i a'ni 'ka (1462 'Nung 3:33a)
‘isn’t it that one knows?’ The postmodifier 'kwo ‘question’, like 'ka, occurs after a noun by omitting
the copula modifier. It lenites after /, y, and i : mu'sum elkwul 'Gwo (1462 'Nung 3:59a) ‘what face is
it?’, mu'sus "ccwoy 'Gwo (1463 Wei 1:7a) ‘what sin is it?’, hon ka'ci a'ni []Gwo (1482 Nam 2:42a)
‘isn’t it the same?’. After other vowels, both elided and unelided velar are found: "es.ten cyen'cho
[G]wo (1482 Kum-sam 3:52b) ‘what kind of cause is it from?’, i mu'su 'kwo (1482 Kum-sam 2:41a)
what is this?’, mu'sum yang'co 'kwo (1462 'Nung 3:84a) ‘what are the looks?’; ne ’y susu'ng i 'nwu
'kwo (1447 Sek 23:41b) ‘who is your master?’; "es.tyey il'hwum i 'PALQ-"ZYA [G]wo (1464 Kumkang
se:8b) ‘why is the name prajna (wisdom)?’ In Chinese passages the Chinese particle YA is followed by
the Hankul '[GJa ( ? 1468- Mong 53a) or '[GJwo (1482 Kum-sam 3:52a).
When attached to a noun that ends in basic -h a metathesis takes place: -h k- -*• ■■■kh The
only example offered for 'ka is hona'kha [= honah 'ka] ye'sus 'ka ‘are they one or are they six’, and
that is given by LCT 706a as “1462 'Nung 106”, but both the locus and the citation seem to be in
error; it was perhaps taken from hona'khwa [= honah 'kwa ] ye'sus 'kwa ‘one and six’ on the
preceding page. The form in the example, however, is just what we would expect, parallel to that of
'kwo in ku "es'te ’n 'sta'khwo [ = stah 'kwo] (1463 Pep 5:165a) ‘what land is that?’.
The 15th-century lenition of velar-initial particles was indicated orthographically throughout the
16th century. But in texts from the 1540s and later (cited by NKW 54-5) there are a number of random
spellings of unlenited 'kwa after vowels (including i), y, l, and -[h], often in close proximity to a
lenited version. The 17th-century texts always write k after l and often (randomly) after a vowel, as
well. For both 'kwom and 'kwos the velar was written in the few examples found in later texts. The
elided form of 'kwa, spelled “wa”, was always used in 19th-century texts (as in the 20th century), but
there was some random variation in the 17th and 18th centuries, as seen in twos.th oy kwoki wa yes
kwa yang uy kwoki kwa (1799 'Nap-yak 18a) ‘the meat of the pig and the meat of the fox and the
sheep’.
There are examples of lexical lenition, as in kaci-Gaci (s nay lol) (1569 Chilqtay 10b) ‘all kinds
(of smells)’, lexicalized from ka'ci ka'ci (1463 Pep 5:137b). And there are velar elisions in nativized
Chinese words:
mwo[k]'ywok (1489 Kup-kan 1:104, 1527 Cahoy 3:11a = 5b) ‘bathing’ < ' MWOK-'YWOK (1447
Sek 6:27b) and 'mwo[k]'ywok thang'co ( ? 1517- Pak 1:52a) ‘bathhouse’ < thang-~co.
'lywu[k]-'we'l uy ( ? 1517- 'No 1:27b) = nywu[k]-wel (1608 Thay-yo 16a) ‘June’ < ' lywuk-
■ngwelq; Cf 'si-'Gwe'l ey n’ ( ? 1517- Pak 1:18a) < ' ssip-'ngwelq ‘October’.
soyng-[k]ang (1527 Cahoy [Tokyo] 1:14a, 1489 Kup-kan 6:21 [cited from LCT]) > soyng-yang
(1562 Cahoy [Hiei] 1:7b, 1583 Kwang-Chen 3b) = soyngkang (1583 Sek-Chen 3b) ‘ginger’ <
SOYNG-KANG.
Although the MK spelling of final unlinked to a following vowel is usually to be treated as
-IG-, when the vowel begins a particle in those texts (such as 1449 Kok) that treat particles as separate
words, no “G” is to be written: i 'nal ay 'za (1449 Kok 109) - Cf mozom 'ay (ibid.). In other texts
we find na'l ay, mozo'm ay, „. , syllabified phonetically. But the -G- is indicated, as part of the noun,
for forms of azolazG- ‘younger brother’ such as az'G on (1445 'Yong 24), az'G i (1445 'Yong 103),
az'G i'la (1462 'Nung 1:86a), and az'G ilwo'n i (1462 'Nung 1:76b).
2.11.2. Labial lenition and elision.
Spellings with ••• G- do not always derive from a lenited velar; some are from lenited labials. We
know this either from variant forms that retain the labial or from modern dialect versions with -p-.
Despite that information, we will write the MK sound as G except when there are morphophonemic
grounds to do otherwise, as there are when other forms of the paradigm of a verb contain a labial. In
the case of e'lwu sol'fWJwo'l i 'syas'ta (1463 Pep 4:70-1) ‘it will be possible to tell them’ we choose
to identify clearly the stem by noting its ellipted W, a lenition of p, as /W] rather than write the G that
would be called for by our rules, because of other forms in the paradigm and also the competing
A Reference Grammar of Korean
PART I 57
version found in sol'Wwo.l i ’ n i (1449 Kok 2). But because of what happened to the vowel (- Wu- >
-Gwu-) we write 'chiGwun (1481 Twusi 6:43a) = 'chiWun (1459 Wei 18:51a) ‘cold’ < *chipu-. The
phoneme G when used represents a neutralization, thought to be the result of merging the labialized
velar fricative and the velarized labial fricative, articulations that are hard to keep apart.
The lone particle is pu'the ‘(starting) from’, shortened from (■■• '%/J pu'the, a verb infinitive.
That particle usually keeps its initial, even after i, y, and /: i pu 'the (1447 Sek 13:1a, 1463 Pep
1:65a) ‘starting from this’, "a'lay pu'the (1449 Kok 109) ‘from earlier’, "nyey pu'the (1459 Wei
2:70a) ‘from long ago’, [no example of •••/?]. There are only two or three examples of a lenited form
Gwu'the < *Wu'the < pu'the, such as wo'nol Gwu'the ( ? 1517- ’No 1:35b) ‘starting from today’ and
"en'cey Gwu'the ( ? 1517- Pak 1:13a) ‘since when’. And there are no examples of *-phu'the with
metathesis of noun-final -h; instead the independent form of the noun appears, as in wu pu'the (1464
Kumkang se:6a) ‘from above’. The modern particles pota and poko are derived from the transferentive
and the gerund of the verb po- ‘look at, see’.
No verb endings begin with p~, but the bound adjective -p- is incorporated in a group of
subjective adjectives that end in -W-, such as "swuyW- (> swiw-/swip-) ‘be easy’ from "swuy- ‘rest’.
And the verb stem pat- ‘butt’ becomes -'Wat- > -Gwat- > -wat- to derive a few intensive stems:
koli'Gwat- ‘conceal’, nilu'Gwat- ‘raise’, ta'Gwat- ‘approach; defy’, mulli'Gwat- ‘spurn; repel’,
penguli'Gwat- ‘crack/split it’, thiGwat- ‘push up against’, wuy'Gwat- ‘lift up’.
Lenition of -p - in compound nouns: tay-'We'm ul (1445 ’Yong 87) ‘a mighty tiger’ < "pern
‘tiger’; phwunglywu-Wa'ci (1459 Wei 24:28b) ‘musician’ < - pa'ci ‘a professional’; kolo-'Wi (1459
Wei 1:36b) ‘a fine rain, a drizzle’ < pi ‘rain’; mwosi 'Gwoy = mwosi 'pwoy ( ? 1517- Pak 1:51b; in
contiguous passages) ‘ramie cloth’ (> mosi pey); pwul'Gwep ( ? 1517- Pak 1:74b) = pwul 'pep (id.
1:75b) < ' pwulq-'PEP ‘Buddha’s Law’, syel'Gwep (id. 1:75a) < ' sywelq-'PEP ‘explaining the
Law’. There is even one example of pye- > -Wye~\ 'swoy-Wye'lo (1445 ’Yong 3:13b [Chinese text])
‘Iron Cliff’ [placename]. Examples that survive in the modern language include si’-o li ‘15 leagues’
and si’-wel ‘October’ < 'si-'Gwe'l ey n’ ( ? 1517- Pak 1:18a) < ’ ssip-'ngwelq ; Cf yu’-wel ‘June’ <
'lywu[k]-'we'l uy ( ? 1517- ’No 1:27b) = nywu[k]-wel (1608 Thay-yo 16a). An elision of noun-final -p
gives us ci[p] s ‘of the house’ in ku cifpj s 'sto'l i (1447 Sek 6:14a) ‘the daughter of that house’, na 'y
cifpj s ke[']s ul ( ? 1517- 'No 2:49a) ‘things of my house’, i'Gwut cifpj s nul'ku'n i (id.) ‘the old man
in the house next door’, ... .
The most common labial lenition is that of the -W- stems, with -p- before a consonant but -W- >
-w- before a vowel. Most of the -w- stems are adjectives, but there are a small number of verb stems,
too (see §8.2.3). The -w- stems include -IW-, for which the only modern example is the literary selw-
= selew- < "syelW- ‘be sad’. MK had "ptelW- = ttelp- ‘be puckery’, "polW- = palp- ‘tread’, and
"yelW- = y e /alp- ‘be thin’, which do not lenite in the modern language, and also "kolW- ‘line up;
compare’, which is obsolete. Other MK -IW- stems: [*']molW- ‘be sad’, "solW- ‘humbly say’
(modern saloy-), "skolW- ‘be difficult’, "tulW- ‘pierce, bore’ (modern ttwulh-, Hamkyeng twulp-).
For a few words we know from dialect evidence that the -G- within an opaque lexical item
represents the weakening of a labial stop, rather than a velar. The noun i'Gwuc ‘neighbor’ (1462
'Nung 3:37a) has the modern dialect versions ipuci, ipucey, iput, iwuci, iwut, as well as the standard
iwus, and it is written with the phonograms “ i-pun ” in 1400+ Cosen-kwan. (The Hankul ip us in 1569
Chiltay is thought to be one of many dialect influences from Kyengsang, where it was published.)
2.11.3. Apical lenition; elisions of / and n.
The apical stop t weakened to the flap allophone of /. That, we assume, is what accounts for the
peculiarities of the -t/l- (or “leniting t”) stems as contrasted with the regular -t- stems. The final -t-
of the leniting stems was replaced by •••/- when a vowel followed.
There are a few etymological examples of lenited t -* l (Martin 1983:27): mwolan ‘peony’ (1576
’Yuhap 1:7b) < "MWUW-tan, cho'lyey (1527 Cahoy l:34a=18b) ‘order’ < 'CHO-'TTYEY (1447 Sek
19:8b). In certain cases the source of the lenition survives as the modern affricate because of the
(southern) merger ty > c, as in the doublet tolyang / tocang ‘Buddhist seminary’ < (?*) "twotyang <
"TTWOW-TTYANG (1447 Sek 24:36a). The MK doublet pa'lol (1445 ’Yong 2) = pa'tah (1459 Wei
58 PART I
A Reference Grammar of Korean
1:23b) ‘sea’ may go back to a hypothetical *pa'talh. And the adverb 'tat (1447 Sek 9:16a) suggests
that talo-ltalG- ‘be different’ < *talok- (§8.3.1) is a lenition of *tatok-. The noun me'li < ma'li
‘head’ is written with the phonograms “ ma-ti ” in 1103 Kyeylim. The Ceycwu word iti ‘this way’
(Seng 'Nakswu 1984:24) appears to preserve the unlenited source of MK i'li. The copula forms ila <
'i'la and iley are probably lenited from ita < i'ta and itey.
Before another apical sound ( t , c, s, n, and rarely / itself) the phoneme / was often elided. The
elision was quite regular in verb forms, so that the word representing "al-'ta ‘knows’ was pronounced
and spelled "a'ta. In listing these -L- verbs we take note of the ellipsis (writing, for example, "a[l]ta)
but in general we follow the spelling and write the forms as they sound. (The headings of dictionary
entries such as “ "alta" in LCT and NKW can be thought of as written morphophonemically or
etymologically.) Though we often note elided consonants such as ~[1J~ with an apostrophe, that is not
done within a stem or paradigmatic form: 'sa'ti < "saflj-'ti, "sa-'two < "sa[l]-'two or < "sa['tij
'two, "sa'nwon < "sa[l]-'nwon. Many modern dialects continue to elide the liquid before an apical,
and even in Seoul the I is sometimes suppressed before -ca (maca) ‘as soon as’ (Part II, -ca Note 1).
In fishing villages of North Kyengsang (Choy Myengok 1979:23) when -1 is stem-final it drops before
ALL consonants, not just apicals, and stems ending in ••■lm- are reanalyzed as -mu-. The dropping of
stem-final •••/- takes place also when stems are compounded: "safl]-'ni- ‘go on living’; "nwo[lJ-[']ni-
> nonil- ‘stroll’; "two[l]-[']ni- > tonil- ‘walk around, circle’; "wu[lj-'ni- ‘go on crying’.
In specific phrases MK nouns ending in •••/ (including the reduction of basic ~lh) sometimes elided
the liquid before the genitive particle s (Cf He Wung 285, 313-4): pa[l] s pata'ng ay (1462 'Nung
10:79a) ‘on the sole of the foot’; kiflhj s ko'z ay s namwo mi'th uy i'sye 'sye (-151 1~ 'No 1:27b)
‘under a tree by the side of the road’. That seems to be the origin of the adnoun mus ‘many, all (sorts
of)’ < mwus < mwu[l] s ‘of the crowd’ and of the noun tokki ‘ax’ < "twos'kwuy < *'two[l]s
'kwuy ‘(ear/)edge of stone’. Spellings that leave the •••/ s or -n s intact tell us the particle was probably
set off by juncture. For an unusual elision before n- of the string /7 s], see Part II, s NOTE 1. Less
often, a noun-final n is elided before the particle s: swofn] s twop (1462 'Nung 1:51b) - preserved by
1874 Putsillo as swotthopi ; i ma[n] s kam yang uy ( ? 1517- 'No 2:22a) ‘for no more sheep than this’
= i man.skan yang ey (1795 'No-cwung [P] 2:20a; by then presumably pronounced Inkkl) = i man
yang ey (id. [K] 2:21a). But the several examples of mafn] 'two may be simply 'ma + 'two: ~
cywungsoyng 'ma 'two "mwot hwo-ngi ’ ta (1447 Sek 6:5a) ‘is inferior to [the life of] any living
creature’. (It is quite possible that the final n of man developed from a separate morpheme attached to
'ma.) See Part II.
In compound nouns final / sometimes drops before n, l, s, or c: pso[lJ-"nwun (1527 Cahoy
1: lb = 2b) = psol-nwun (1576 'Yuhap 1:4a) ‘pellets of dry snow’ (ssalaki nwun); chil-'pha[l] li s
kil'h i ( ? 1517- 'No 1:60a; < chilq-'palq ~li) ‘a road of seven or eight leagues’. The word pha’-il
‘the eighth (of April = Buddha’s birthday)’ must have elided the I of phal-il < PALQ-'ZO.Q at a time
when the Chinese morpheme for ‘day’ was felt to begin with an apical sibilant.
Etymologies with elision of l before an apical (LKM 1963:87-8) include:
kyeo-sali (1748 Tongmun 2-46; cited from LCT) ‘mistletoe’ < *kyezu(l) sali (‘winter life’);
“wiNTER-u/-^a-//” (1431 Wellyeng), “winter-m-; ” (1250 Hyang-kup);
kwufl]-cwokay ( ? 1544- Akcang: Cheng-san) ‘oysters and clams; oyster (with shell)’ = kwul s
cywokay (1489 Kup-kan 6:81; cited from LCT); “ kwulq-cwo-kay" (1250 Hyang-kup);
minali < *munali < *mul nali ‘parsley’; “water -nay-lip” (1250 Hyang-kup).
In a maximally informative notation all elisions can be shown with the elided consonant between
brackets, as above for [l]. And it is possible, as set forth in Part II, to treat -flj-'two as -f-tij 'two
though that is probably not the best historical explanation. When the elision is between two words (or
within a noun compound), the apostrophe is used: kau’ nay ‘throughout the autumn’, kyewu’ nay ‘all
A Reference Grammar of Korean
PART I 59
winter long’; atu’ nim, tta’ nim; so’ namu ‘pine tree’; cha’-tol ‘quartz; silicon’; 'na’-'tol ‘days and
months’; ppu’-takw(un)i ‘a part or corner sticking up’ (< ppul ‘horn’); mu’-soy ‘cast iron’; panu’ dl
‘needlework, sewing’, kantu’-cak kantu’-cak ‘swaying gently’ (< kantul kantul); cha’-pssal
‘(hulled) glutinous rice’.
Before i and y the phoneme n is sometimes weakened to just nasality and even dropped. Ramsey
(1978:52-3) says that in South Hamkyeng both n and ng are reduced to nasality on the preceding vowel
when y or i follows, but a trace of the apical articulation of the n remains in that the preceding vowel
will not be fronted (in the usual assimilation to the following y or i). Speakers in various places say
[ai] or [ai] for ani ‘no’. Examples of n-elision from fishing villages of North Kyengsang (Choy
Myengok 1979:23): ku key cip i a(y)ila(y) ‘that’s not a house’; mad = manh.i ‘lots’. The MK noun
"kwoy ‘cat’ has the earmarks of a missing consonant *kwo'[]i and 1103 Kyeylim writes the word with
the phonograms “ kwo-ni ”. Quite a few modern dialects, including that of Ceycwu, attest the word as
konayngi (Kim Hyengkyu 1974:b-170), retaining the nasal.
2.11.4. Sibilant lenition and elision.
Before a vowel the stem-final consonant of -(s)- verbs (§8.2.5) like cls-/ci- ‘build’ and nas-/na-
‘be/get better’ vanishes in modern Seoul, but in MK it was a voiced [-z-], so that the infinitives were
spelled ci'ze and na ze, using the obsolete triangle symbol for the MK Izl. Since modern dialects retain
forms such as /nasa/ for naa ‘get/be better’ *- na(s)-, we assume that the MK version was an
ephemeral lenition of an earlier -s-, a lenition which took place under conditions absent from the
regular s-final consonant stems (see Martin 1973, Ramsey 1975). Most of the modern -s- stems were
leniting in the 15th century, including as- ‘seize’, but pes- ‘take off (clothes, ... )’ and its obsolete
variant pas- are attested only as regular •s- stems. And MK is- ( > modern iss-) ‘exist’, a contraction
of MK isi-, never lenited, for that allomorph was used only before consonants. The verb wus- ‘laugh’
was a leniting stem in the 15th century, with the infinitive wu'ze, but the lenition did not survive to
give us modern *wue, and the stem is regular today. Kim Wancin (1973) believes that this stem was
restructured because of a clash between the homonyms that developed for such forms as MK wu'lun >
wun ‘ that cried’ and wu'zun > wu-un -* wun ‘ - that laughed’; but notice that 1876 Kakok has
(122) wu-un and (90) wu-um as forms of ‘laugh’. Are there any modern dialects that treat the verb as
leniting?
According to the map in LSN 1956:103 the unvoiced [s] is retained for MK z in dialects of the
northeast and much of the south but is lost in the middle and the northwest; the area where -z- was lost
is somewhat wider than that of the loss of distinctive pitch accent but it covers that territory. LSN
concludes that the assimilatory voicing of -s- between voiced sounds (vowels, l, m, n) arose in the
middle and northwestern areas during the middle ages (1300-1600); it was apparently not present in the
Sinla language, he says, and it began to disappear again in the middle of the 16th century. He did not
address the question why some MK words have -z- but others have -s-. Unless we can establish
prosodic conditions for the lenition, the exceptions will have to be treated as the result of dialect
mixture or the failure to maintain or even establish an orthographic tradition in the midst of the
collapse of the distinction. Did the modern dialects lacking the distinction of -(s)- stems ever go
through a stage when they had the voiced [zj? Probably not, though that would be implied by an
explanation that would have their modern paradigms restructured by analogy. As mentioned in §2.4, it
is rare to hear [z] as an allophone of /si in any modern dialect, but there is evidence for that in earlier
Hamkyeng speech as reported by Putsillo in 1874 (see Martin ? 1991, King 1990), though it is
apparently absent in that area today. There would seem to be no good motivation for a voiced fricative
sound to become voiceless precisely in all-voiced environments, so I assume that modern s (rather than
elided [z]) in areas for which the voiced version was earlier reported must be due to people being
overwhelmed by the habits of nearby speakers who never gave into the lenition.
Some of the MK words with -z- appear with s in modern dialects: kozolh/kozol ‘autumn’ >
kasil, kusul, kisil as well as standard kaul; mozolh ‘village’ > masul, masil, maswul, ... , as well as
60 PART I
A Reference Grammar of Korean
standard maul; mwuzu / mwuzG- < *musuk ‘radish’ > musu, musi, muku, mukkwu, mutkwu, and
(Pukcheng) mukk as well as standard mwuwu. The word for ‘kitchen’ appears both as puzep (1451,
1466, 1481) > puep (1632) and as puzek (1481, 1527) > puek (1632) and has the modern dialect
versions pusep, pusap, pusek, pusik, pucek, pudk, (Pukcheng) pekk as well as standard puekh
(usually treated as puek in Seoul); it is probably from an old compound with pu[l] s ••• ‘of the fire’
attached to a variant of seph < *s(y)ep[h] (the syep spelling is in 1632 Kalyey but earlier attestations
are all sep, and the hypothetical final h is not attested before the 20th century), perhaps a variant of
nyeph/nyekh ‘side’, which would help account for the doublet forms with -p and -k. It is unclear how
far back the final aspirated velar of the standard version puekh can be traced; Scott wrote pwuek ey
(1887 Scott 196 = 1893 Scott 240). For more on puekh see Ramsey 1984. The noun mozom ‘heart,
mind’ > maum is reported as mourn or mosum in Ceycwu dialects; the MK z may have lenited from
c, since phonograms of 1400+ Kwan-yek (item 405) represent “mo-co/n”. It should be kept in mind
that a few of the -z- words survived with an affricate (instead of s or zero): honca ‘alone’ (see the
entry in Part II and the discussion in §2.10.4).And the particle kkad < 5 koc(i) < s ko'c(ang), as
well as ko'cang ‘end’ and (> kacang) ‘most, very’, offers evidence for the history of "kos < ko'zo
< *ko'co ‘brink, edge’. LKM 1972a:38-9 calls attention to the forms 'swon'cwo ( ? 1517- Pak 1:63a)
< 'swonzwo (1447 Sek 6:5a) ‘by his/her own hands, personally’ and mwomcywo (1617 Sin-Sok
chwung 1:36) = 'mwomswo (1586 Sohak 6:25a) [= chin hi 1518 Sohak-cho 9:27b] < 'mwomzwo
(1481 Twusi 6:34a) ‘with one’s own body, personally’, and would explain modern samcil ‘Double
Three day (= the third of March)’ as continuing a nativized SAM-'ZHQ ‘day three’. He also mentions
MK namcin ‘man, husband’, which must be from a nativized NAM-ZIN.
Some of the -z- words are lexicalized from phrases: "ilza'ma (1462 'Nung 6:70b, 1463 Pep
7:159a) ‘indulging in’ < "il flwo) sa'ma ‘making it one’s business’. And "twu'zeh ‘a few’ is from
" twu[lh j + a variant of "seyh ‘three’. Other examples of lenition of a morpheme-initial sibilant include
the deferential - "zow- and the particle 'za.
The stem "wuzW- ‘be laughable’, found in "wuzWu'l i (1445 'Yong 16) and "wuz'Wi (1449 Kok
179) has undergone a second lenition (labial) by attaching the leniting bound adjective -p-\ the source
form must have been *wusupu-. The stem "yez'Gwo- (" yez'Gwa 1462 'Nung 10:41a, "yes'Gwa 9:87b)
derives from *yez-Wwo- < "yes-'pwo- ‘spy on’ (He Wung 126).
2.12. The accent of earlier forms.
The language spoken in modern Seoul differs from the 15th-century language in lacking accentual
distinctions that are still found in northeastern and southern parts of the peninsula in the form of
patterns of pitch or combinations of pitch and vowel length. Although some of the southernmost
dialects (such as Kimhay) have three pitch levels - high, mid, low - others (such as Antong) have
two levels, high and low, but distinguish some of the low syllables by lengthening the vowel. Still
other dialects (such as Hamhung) merge those two categories but end up with both pronounced high
and the expected high pronounced low. And in older Seoul speech the vowel length of the third
category was preserved, but a difference of pitch level got lost, so that there was a merger between the
short high and low syllables. In most of the texts of the 15th century a single dot was placed to the left
of a high-pitched syllable, a double dot to the left of syllables that were long and rose from low to
high, and the low short syllables were left unmarked. When a word is cited in isolation without tone
marks we cannot be sure whether it represents all low syllables in a tone-marked text or is taken from
a text that did not mark the tones. In words and phrases of more than one syllable the stretches of
tones formed accentual patterns, much like those of Japanese pitch accent, and that makes Korean
different from a tone language such as Vietnamese or classical Chinese. Scholars of the early Hankul
period were acutely aware of the traditional Chinese four tones (sa-seng) and wrote these for the
Hankul readings of Chinese characters, equating the low tone with the “even” tone (phyeng-seng), the
high tone with the “going” tone (ke-seng), and the long low-high with the “rising” tone (sang-seng).
A Reference Grammar of Korean
part i 61
The “entering” tone (ip-seng) was posited by Chinese phonologists to account for the syllables ending
in unreleased voiceless stops ~p, -t, and -k, still heard in Cantonese but lost in northern Chinese. In
borrowing Chinese words, Koreans treated those stops as -p, •••/, -k, with the apical version written
■■■LQ in Hankul readings of characters, and marked them with the single dot of the “going” tone. But
the character readings, with respect to tone as well as other features, were somewhat artificial,
constructed to conform to the information that Chinese phonologists had compiled as riming guides for
the language of the 7th century. Chinese words that got into popular usage were often treated
differently, though the “even” tone (accounting for almost half the Chinese morphemes) was usually
equated with the Korean unmarked low tone.
The accentual patterns of native Korean words did not, of course, come from China, but must be
considered a distinctive part of the ancestral forms. Because of partial correlations between the accent
and the canonical shapes of morphemes, it is suspected that at least some of the patterns were internal
developments, so that possibly the ancestral language may have treated pitch as nondistinctive; but that
hypothesis remains unproven. The low-high rise marked by the double dot is often the result of
collapsing two syllables (low + high) into a single long syllable; and sometimes, especially in verb
forms, it indicates retention of the high pitch of a syllable that elided its vowel, typically the
nondescript u or o that represented a minimal vowel quality. Thus for ‘walk’ the modifier ke'lun <
*ke'tu-n (1459 Wei 1:27b) has an overt vowel to carry the high tone, but the gerund "ket'kwo <
*ke't[uJ-'kwo (1449 Kok 130) does not, so the high pitch is added on to the low pitch of the preceding
syllable. (See Ramsey 1978a:209-24.)
The MK tone dots have been transcribed as they appear (or do not appear) in the cited passage
when that is from a photocopy of the original text or when it is from a secondary source that included
the information. The examples from 1445 'Yong carry the accents of LKM’s 1962 interpretation. I
have added in brackets a few tone marks that I think are missing because of broken type, or that are
the result of surface processes (such as the frequent loss of a dot from a string of three dotted
syllables), where the restoration helps the reader see the structure. I have not, in general, given the
basic or reconstructed dots for strings of morphemes in endings. There, as toward the end of long
phrases in general, a tendency developed to disregard distinctive accent after the first in a stem (or a
noun), either omitting the marks or indicating an automatic “sing-song” tune of alternating pitches.
The tendency became quite noticeable in 1481 Twusi, where Kim Chakyun (1979), like Kim Wancin
and Ceng Yenchan earlier, observes that many particles and endings that had been marked ’ (high) in
1445 'Yong, 1447 Sek, 1449 Kok, and 1459 Wei are left unmarked (low) or marked " (low-high), and
that is even more striking in 1587 Sohak; he characterizes the trend as (38) “neutralization at end of
word”. We have left unmarked the frequent suppression of a final dot on the infinitive ending when a
particle (with dot) is attached, such as e'wule 'za (1451 Hwun-en 13a) Cf e'wu'le (1463 Pep 2:114b).
And for the most part we do not call attention to suppression of the second of three dots on contiguous
syllables, as in 'pwola'm ol ( ? 1468* Mong 20a) ‘the sign’ -Cf 'pwo'la.m in i (1459 Wei 21:217-8),
but we make a few exceptions when it helps account for the words in a phrase: 'wos 'kwa 'pap [Jkwa
'ay (1481 Twusi 16:19a), 'swoy [Jyey 'sye (1459 Wei 2:28b), swo'li [']yey 'sye (1447 Sek 24:1b),
hoy [ Jyey 'za (1447 Sek 23:13a), i kak'si [Jlwo 'za (1459 Wei 7:15b). We leave unmarked the
suppression of an accent in "twuy.h ey nun (1445 'Yong 30) = "twuy[']h ey nun, Cf "twuy'h ey n’
(id. 70).
There are numerous examples of the crasis of a final low pitch with the tonal residue of an ellipted
high-pitch syllable (Cf He Wung 337): "nay'h ay to'li "eptwo'ta (1481 Twusi 25:7a) ‘there is no
bridge on the river’ < toli [ij; ~co 'non mot nwu"uy ’'Gwo 'MOY 'non azo nwu"uy ’ la (1459 Wei
21:162a) ‘[the Chinese word] "co is an older sister, moy is a younger sister’ < nwuuy i'Gwo -
nwuuy 'i'la. That is what accounts for the rising accent in preconsonantal forms of the -L- stems
(§8.3.2).
62 PART I
A Reference Grammar of Korean
2.12.1. Accentual patterns.
A limited number of patterns were available for words of a given length. The patterns are shown
below, with examples, for nouns and adverbs of one, two, and three syllables.
L
H
R
mol ‘horse’
mal ‘measure’
"mal ‘words’
LL
HL
LH
LR < LLH
RL
pwoli ‘barley’
'sell ‘midst’
me'li ‘head’
mak"tay ‘club’
"cyepi ‘swallow’
HH(/HL)
RH
'mwo 'koy ‘mosquito’
"ke 'cus ‘false’
(HLL/)HLH
LHH
LLH
RHH
twos.ka'pi ‘goblin’
mye nu li ‘wife’
kama 'kwoy ‘crow’
"a 'mo 'lyey ‘how
HLH
LHL
LLR
RLH
muci'key ‘rainbow’
ye'tolay ‘8 days’
(")
"soma kwoy ‘mo
LLL
HHH
RLL
cintolGwuy ‘azalea’
kuy lye ki ‘goose’
(-)
LRL
HRH
HLR
LHR
RHL
(-)
(")
ema "nim ‘mother’
a pa "nim ‘father’
"ke'cu-'mal ‘lie’
RRL
RRH
RLR
(-)
(-)
(?)
Ramsey, following Ceng Yenchan (1971), assumes that certain patterns freely varied with each other:
HH / HL; LHL / LHH; HLL / HLH / HHL. That assumption is made for two reasons: the patterns are
merged in the reflexes found in the modern dialects, and for some of the words the early attestations
vary. We find three kinds of evidence for a given etymon:
(1) Only one pattern is attested. For these, we have no direct evidence that the pattern was not
distinctive.
(2) Two or three patterns are attested and the variants are unmotivated, in that they cannot be
explained by their environments. For certain words there may be only one attestation for a variant
pattern, while for other words there are several attestations for each pattern.
(3) Certain words accentuated H(H)H are converted to H(H)L when a particle or copula expression is
attached. This is a prosodic adjustment, as if to avoid a long string of high syllables, though such
plateaus can be found in other phrases.
In the examples that follow, the English gloss is given first.
Type H X X
HHH
‘wild goose’ kuy'lye'ki (1527 Cahoy 1:8b = 15a)
HLH / HHL
‘goblin’ 'twos.ka'pi (LCT; 1447 Sek 9:36b, 1449 Kok 163, 1482 Kum-sam 4:23a); twos'kapi ’n i
(1459 Wei 21:105a).
HHH / HLH (? or HLL, attributing the accent to the particle)
‘crane’ 'twu'lwu'mi (LCT; 1459 Wei 7:66a, 1527 Cahoy 1:9a= 16a); twulwu mfij uy ( ? 1517- Pak
1:27b)
HLH
‘rainbow’ muci'key (1445 'Yong 50, 1462 *Nung 2:87b, 1481 Twusi 16:42b, 1527 Cahoy l:lb = 3a)
‘tadpole’ wolchang-'i (1527 Cahoy l:12b=24a) = 'wolchang (1446 Hwun 29a; the dot is strangely
missing in the Taycey-kak repro, but it is clear on the photo plate included in *Yi Sangpayk 1957)
‘rather’ wohi'lye (1459 Wei 1:37a, 21:149a; 1462 'Nung 2:67a; 1463 Pep 2:77a, 2:158a, 4:192-3;
1475 Nay 2:1:2b)
A Reference Grammar of Korean
PART I 63
HLR
‘mother’ 'ema "nim (1445 'Yong 90, 1449 Kok 16, 1459 Wei 2:6b), ~ i (1459 Wei 8:84b) = ema
"ni'm i (1459 Wei 21:27b, 21:28a)
HLH / HHH
‘grandfather’ 'hana'pi (= ha.n a'pi) (1527 Cahoy l:16a=33a), 'hana'pi V (1445 'Yong 125);
hana'pi ( ? 1517- 'No 2:34a)
HHL / HLH / HHH
‘peak’ 'tyeng'paki (1447 Sek 6:43b, 1459 Wei 18:16b); 'tyengpa'ki (1465 Wen 2:2:1:38a);
tyeng'pa ki (1527 Cahoy l:13a=24b) ‘peak’
HLH / HHL
‘elephant’ kwokhi'li ’Gwo (1459 Wei 1:27b), kwokhi'li ’n i (id. 1:28a); kwo'khili (1527 Cahoy
1:9b = 18a) < *'kwoh ki'lfu]-'i ‘nose long-one’ = ‘long-nosed one’
Type L X X
LLH
‘ant’ kayya'mi (1447 Sek 6:35 ['Yi Tonglim version], 6:37, 1481 Twusi 15:56a, 1482 Nam 2:32a),
kayya'mi /’ (1481 Twusi 7:18b) = ka.ya'mi (1459 Wei 18:39b, 1481 Twusi 8:8a, 1482 Kum-sam
5:36a)
‘branch’ kayya'ci (1481 Twusi 23:23a) = ka.ya'ci (1481 Twusi 10:5b)
‘here’ inge'kuy (1447 Sek 19:17b; 1459 Wei 13:35b, 14:59a; 1481 Twusi 7:14a; 1482 Nam 1:14a)
LHL
‘eight days’ ye'tolay (1459 Wei 2:35ab)
‘mullet’ ka'mwothi (1527 Cahoy l:lla=20b)
LHR
‘acorn’ two'thwo-'pa.m ol (1481 Twusi 24:39a) = LHH two'thwol-'Gwam (id. 25:26b)
‘father’ a pa "nim (1449 Kok 23); ~ ul 1459 Wei 8:84b, ~ 7 10:2a; a pa "nim s 'kuy (1447 Sek
6:1a)
LHH
‘harp’ ke'mun-'kwo (1481 Twusi 16:30b; 1586 Sohak 6:94b), ke'mun-'kwo y (1481 Twusi 21:35a)
‘nine days’ a'ho'lay (1447 Sek 9:31a = 1459 Wei 9:5lab)
‘puddle’ wung'te'ngi (1527 Cahoy l:3a=5b)
‘sneeze’ co'choy'ywom (1475 Nay 1:49b, 1527 Cahoy l:15b=29b, 1586 Sohak 2:7a)
‘wife’ mye'nu'li (1447 Sek 6:7a, 1527 Cahoy 1:16a=3lb, 2:1a)
‘acorn’ (1) two'thwo'li ([1517—»•] 1614 Saseng 2:68a [dots obscured in repro], 1527 Cahoy 1:11b)
(2) two'thwol-'Gwam (1481 Twusi 25:26b)= LHR two'thwo-'pa.m ol (id. 24:39a)
LHL / LHH
‘aunt’ a'comi (1445 'Yong 99, 1481 Twusi 8:38a); a'co'mi (1527 Cahoy l:16b=31b, 1:16b = 32a),
a comfy] oy 'swon-toy (1475 Nay 2:1:29b)
‘midst’ ka'won-toy (1482 Kum-sam 2:65a), ka'won-toy s (1482 Kum-sam 2:31b); ka'won-'toy (1447
Sek 6:31a; 1459 Wei 1:4a, 1:30a, 2:51b, 9:22b; 1462 'Nung 2:84b, 3:38a; ?1468- Mong 43b, 64b;
1481 Twusi 15:44a; 1518 Sohak-cho 8:32b), ka'won-'toy n’ (1449 Kok 70), ka'won-'toy V (1482
Kum-sam 2:65a), ka'won-'toy s (1527 Cahoy 3:34b = 15b)
‘a mute’ pe'weli (1550 'Yenghem lib); pe'we'li (1447 Sek 19:6b, 1459 Wei 17:52a, 1527 Cahoy
2:16b=34a)
‘packsaddle’ ki'luma (1481 Twusi 15:1b, 21:22b, 22:8b; 1586 Sohak 5:54a); ki'lu'ma (1527 Cahoy
2:13b=27a)
‘seagull’ kol'myeki (1481 Twusi 7:37a, 10:2a, 15:53a, 21:38a); kol'mye'ki (1527 Cahoy 1:9a= 16b)
‘traveler’ na'kunay (1481 Twusi 7:2a [LCT is wrong], 7:9a, 7:14b, 7:26b, 10:2b, 15:23a, 15:31b);
na'ku'nay ’’ni ^1517- 'No 1:18b)
At least one noun has variants starting either low or high:
64 PART I
A Reference Grammar of Korean
LLH / HLH
‘cricket’ moyya'mi (1527 Cahoy l:12a=22b), moyya'mfi] oy (1482 Nam 2:40b); 'moyya'mi (1481
Twusi 15:27b; LCT “ 'moyGa'mi” , butya is clear in the repro), 'moyya'mfijoy (1481 Twusi 20:8b)
There are a few words with more than one rise. They are reduplications, phrases, or half-
assimilated Chinese loans:
RR
‘always, ever’ "nay "nay (1445 'Yong 16, 1447 Sek se:2a, 1463 Pep 2:20b [dots obscured in repro])
‘filial devotion’ "hywo'yang (1586 Sohak 6:50b = 1518 Sohak-cho 9:55a) < ' hywow-'YANG
‘grudge’ "wen'mang (1586 Sohak 6:83b) = "wen'mang (1518 Sohak-cho 9:90a) < qwen-'MANG
RRH
‘bird beak’ "say "pwu'li (1527 Cahoy 3:3b = 6b)
To the examples of rise patterns we can add various forms of verbs and compound verbs, taken
from examples in Part II and here listed without gloss or source:
rrl "ti-'nayGwo
rrh "ket-"nay'ya
RLR ?
rll "etusil, "wulGenul
rlh "ayGwa'thye, "azop'kwo, "cyektwo'ta, "cywokwo'ma, "ep.su'sya, "hoyGi'ta, "hoyGwo'ta,
"ilGe'tun, "kyesi'ta, "mantha'la, ", salGe'na , "samke'nul, "sitcop'key, "twolo'sya, "wulGwe'le
rhh "al'Gwa'tye, "cwos'soWa'n, "cye'ku'na, "nam'to'lwok, "sa'ni'ta, "wu'ni'ta
rhl "a'losil, "a'losyam, "a'molyey, "ep'kesin, "ep'siGwul, "ep'susil, "kye'siken, "mey'zoWa
llr kolo "chywom
lrl cap'sopke'n, mas-'nala, tut "copkwo, tu'zopta
lrh tut'cop'kuy, nip"sop'kwo, tut"cop'ti
lhr kol hoy- "nay
hrh ?
Examples of (invariant) LH:
‘butterfly’ na'poy (1462 'Nung 7:83b; 1481 Twusi 15:11b, 21:6b; 1527 Cahoy l:llb=21b), napoy
lol (1463/4 Yeng 1:22b; dot obscured?), na'pwoy (1481 Twusi 15:32a, 23:20a) # nap oy ‘of the
monkey’ (1465 Wen se:64a, 1482 Kum-sam 2:44b) = nap 'uy (1481 Twusi 20:21a)
‘bug ' pel 'Gey (1447 Sek 9:24b, 1449 Kok 28a, 1459 Wei 9:26a, 1462 'Nung 7:83b, 1463 Pep 2:107a
[dot obscured in repro], 1527 Cahoy 3:2a = 3a)
‘fault’ he'mul (1459 Wei 2:6a; 1462 'Nung 4:53a, 4:122a; 1463 Pep 4:36b; 1527 Cahoy 2:17a = 35a,
3:29a); he'mu l (~ ul 1445 ‘Yong 119; ~ i 1447 Sek 9:4b, ~ un 1462 'Nung 7:85a)
‘fish; flesh, meat’ kwo'ki (1447 Sek 6:10b; 1481 Twusi 7:5a, 7:7b, 10:31b, 16:19b, 22:7b; ?1517- 'No
1:22a; 1527 Cahoy 2:1 la=21b, 3:2a = 3a), ~ lol (1447 Sek 9:12a), ~ lol (1481 Twusi 25:14b),
~ non (1481 Twusi 16:19b), ~ 'la 'two (1447 Sek 9:13a); kwo'kfij oy (1459 Wei 1:14a), ~ lan
(1481 Twusi 21:3a)
‘root’ pwul'hwuy (1445 'Yong 2, 1462 'Nung 2:22a, 1463 Pep 2:131a, 1481 Twusi 7:23b, 1527 Cahoy
3:2a = 3b), ~ ''la (1459 Wei se:21a), ~ lol (1447 Sek 6:30b, 1449 Kok 99)
‘scales (offish, etc)’ pi'nul (1527 Cahoy 3:2a = 3a; ~ 'ul 1449 Kok 28a, ~ 'Gwa 1459 Wei 7:35a),
pi'nu'l ey (1447 Sek 13:8a), pi'nu.l ol (1482 Nam 1:64a), pi'nol (1481 Twusi 25:14b)
2.12.2. Accentual variants.
Certain words are attested with two or more accentual patterns. In the most common type, the
variant loses all dots but the first, exemplifying the tendency to neutralize pitch distinctions in the later
part of a word:
‘as if’ 'ma'chi (1447 Sek 6:25b, ?1517- Pak 1:23a, ?1517- 'No 2:66a) > 'machi (1481 Twusi 7:7b,
10:9a; 1482 Kum-sam 2:53a)
‘fitting’ 'mas'tang (1462 ‘Nung 1:89a; error?) > 'mas.tang (1447 Sek 13:12b, 1462 'Nung 10:42b,
1463 Pep 3:196b, ?1468- Mong 20a, 1481 Twusi 8:6b, 1482 Kum-sam 2:37a, 1475 Nay 2:1:49a)
A Reference Grammar of Korean
PART I 65
‘deliberately’ kwu'thuy'ye (1459 Wei 9:13b, 1463 Pep 2:203a, 1481 Twusi 20:29a) > kwu'thuyye
(1449 Kok 145; 1481 Twusi 10:12a, 25:29a)
‘first’ pi'lu'se (1463 Pep 1:131a, 1465 Wen 1:2:3:6a) > piluse (1464 Kumkang se:6b; 1465 Wen
1:2:2:140a, 2:3:1:25a, 2:3:1:52a, 2:3:2:68a; 1475 Nay 2:l:16ab; 1482 Kum-sam 2:3a, 4:36b; 1482
Nam 1:33b)
‘mirror’ ke'wu'lwu (1462 'Nung 2:17b,b; 1481 Twusi 21:35b) > ke'wulwu (1447 Sek 24:20b; 1459
Wei 8:20b; 1482 Kum-sam 2:63a, 3:31a)
‘obligatorily’ mwo'lwo'may (1447 Sek 6:2b; 1451 Hwun-en 13a; 1459 Wei se:17a, 7:15b, 14:31b,
23:91b; 1462 'Nung 4:77a; 1465 Wen 2:3:2:44a; ?1468- Mong 10b; 1463 Pep 4:148b; ?1517- 'No
2:44a) > mwo'lwomay (1475 Nay 1:76b; 1481 Twusi 15:6a, 15:42b, 20:4b; 1482 Nam 1:24a)
‘necessarily’ pan'to'ki (1462 'Nung 1:17a, 1:67a) > pan'toki (1465 Wen 1:1:1:63a, ? 1468- Mong 13a)
‘king’ "nim-'kum (1445 'Yong 33, 49, 84, 121; 1459 Wei 1:31b) > "nim-kum (1481 Twusi 10:9b,
22:46 [faint]; 1475 Nay 1:9b)
‘woman’ "kye'cip (1463 Pep 2:28b; 1447 Sek 6:6b) > "kyecip (1447 Sek 6:4a, 19:19b, 24:2b; 1459
Wei 7:10b, 8:94b; 1463 Pep 4:176a,b)
The honorific term for ‘(one’s) words’ is usually "mal-ssom but when it is before the particle
'o'lwo it appears as "mal-'ssom, as in 1447 Sek 13:48a (" mal-'sso.m o'lwo) and 1465 Wen se:lla
("mal-'so.m o'lwo); Cf He Wung 328. (He Wung misreads 1451 Hwun-en la as a similar example
with the nominative particle, but the text has "mal-sso.m i'la.)
The word for ‘cloud’ is 'kwulwum (1445 'Yong 42; 1447 Sek 19:41b; 1449 Kok 81; 1459 Wei
2:51b, 7:35a, 7:31-2; 1462 'Nung 4:6a, 8:50b; 1463 Pep 3:35a; 1465 Wen 1:1:10b; 1481 Twusi
7:23b, 8:11b, 15:9a, 15:9b, 21:7b, 21:14b, 21:22b, 21:41b; 1482 Nam 1:34a; 1482 Kum-sam 3:36b;
... ) but there is at least one example of ' kwu'lwum : 'kwu'lwu.m i'la (1459 Wei se:18a), yet on the next
page (18b) 'kwulwu'm ulh.
A word for ‘branch’ is cited as 'kaci by LCT and as 'ka'ci by NKW, and He Wung gives a single
example of the latter, ka'ci ’Iwo'ta (1481 Twusi 7:1a), but in the only reproduction of the passage I
have seen the marking is unclear; in any event, it could be treated as ka'c’ ilwo'ta with the second dot
going with the copula form ('ilwo'ta). Other examples (1447 Sek 13:47a, 1449 Kok 19, 1459 Wei
1:43b, 1481 Twusi 8:3b, 15:4a) all seem to be 'kaci. Compare the LH word ka'ci ‘kind, sort’ (as noun
1465 Wen 1:1:2:61a; as counter 1447 Sek 6:4a and 24:2b, 1459 Wei 21:88-9, 1462 'Nung 2:17a, 1481
Twusi 8:24b, 1482 Kum-sam 4:40a), whence hon ka'ci ‘same’ (1447 Sek 13:29a, 23:4a; 1459 Wei
2:61a, 8:31b, 9:22a; 1462 'Nung 1:17a, 1:99a, 2:19a, 6:54a; 1482 Kum-sam 4:20b; ?1517- Pak 1:72).
Ramsey 1978a: 109 has a list of forty two-syllable nouns said to be High-Low or High-High in
Middle Korean, including the above three. But for these three, at least, the attestations of the High-
High versions are very few, as we see above, and perhaps are to be accounted for as a back-shift of the
initial accent of the copula forms i'la and 'ilwo'ta and of the particle 'o'lwo, which often functions as
an adverbialization of the copula (‘so as to be, being, as’). Of the other words listed by Ramsey, for
‘fly’ LCT has pho'li (1527 Cahoy l:llb=21b) and NKW pholi (1481 Twusi 10:28b, 20:26a), but
the High-High pattern occurs only isolated in the 16th-century dictionary; the word is one of several
early examples of the accreted noun /suffix -i and was earlier (1446 Hwun 27b) just phol. The only
tone-marked examples of 'kwuki ‘ladle’ (1527 Cahoy 2:7a=lla), pak.ha ‘mint’ (1527 Cahoy
l:8a= 15a), and toypha ‘plane’ (1527 Cahoy 2:8b= 16b) are High-Low, as are those of:
‘chick’ piywuk (1446 Hwun 28a)
‘owl’ 'puheng (1446 Hwun 28a) = puhweng-i (1527 Cahoy 1:8b = 15b), probably < puhweng
(pa'hwoy) ‘Phoenix (Rock)’ (1445 'Yong text 5:27b) = "pwonghwang (1527 Cahoy l:8a=15a) <
“ PWONG-HWANG
‘midst’ seli ('seli 'yey 1445 'Yong 4, 1449 Kok 124; 1459 Wei 9:35f; 1481 Twusi 7:10b, 10:13a,
16:39a; seli ’ la 1459 Wei 1:19b). If not a mistake, kwulwum se'li Iwo 'sye (1481 Twusi 22:21b)
treats the first two words as a compound; we expect a dot on the first syllable of kwulwum ‘cloud’,
‘taro’ 'thwolan (1481 Twusi 7:21b, 22:56a; 1527 Cahoy l:7b=14a)
66 PART I
A Reference Grammar of Korean
The only attestations of the following words are High-High:
•jar’ tan ti (’1517- Pak 1:41a, 1527 Cahoy 2:7a=12b)
‘rice wash-water’ 'stu'mul (1459 Wei 21:110b, 1527 Cahoy 3:5b=lla)
‘strawberry’ ptal'ki (1527 Cahoy 1:6b = 12a; correct the heading “'ptalki ” in LCT 199a)
‘belch’ thu'lim (1527 Cahoy l:15a=29b)
‘helmet’ 'thwu'kwu (1445 'Yong 52, 89; 1527 Cahoy 2:14a=28a)
‘goat’ 'yem'sywo (1527 Cahoy 1:10a= 19a)
The word for ‘wave’ is normally treated as a phrase: mul s 'kyel (1465 Wen 1:2:3:22b) = mu[l]
s 'kyel (1527 Cahoy l:2b=4b), mul s kye.l i (1449 Kok 107; 1482 Nam 2:58a, 2:58b), mul s 'kyel s
(1447 Sek 13:9b, 1459 Wei 9:22b, 1462 'Nung 8:84a, 1463 Pep 1:51a), 'mul s 'kye.l ul (1463 Pep
1:51a, 1465Wen 1:2:1:28a), mul s kye.l uy (1481 Twusi 8:11b), mul s kye.l i'la (1462 'Nung
1:64a). But the second dot is absent in: 'mul s kye l ul (1481 Twusi 10:3b) and mul s kyel tywung 'ey
(’1468* Mong 43a) - yet mul s kye'l ey (ibid.). Cf ‘tear’ 'nwun s mul (1449 Kok 45, 1475 Nay
2:2:13b; - ['] Gwos 1481 Twusi 8:30a), 'nwun s mu.l ey (1481 Twusi 8:45b), nwun s mu.l ul (1481
Twusi 7:10b) = nwun z mu l ul (1445 ’Yong 91). Other words treated as phrases include:
pal s tung ‘heel’ (~ 'kwa 1463 Pep 2:12a, ~ ul 1463 Pep 1:55a, ~ i'Gwo 1463 Pep 1:55a),
pal s tung i (1459 Wei 2:40b, 2:57a; 1463 Pep 2:12a); and the lexicalized all-low version pa[l] s tung
(1527 Cahoy l:15a=29a).
pal spa'tang ‘sole of foot’ (~ s 1459 Wei 2:37b), 'pal spa'ta'ng i (1459 Wei 2:40a, 1462 'Nung
2:115b), pal s pa'ta'ng ay (1466 Kup 1:32b; 1:63b lacks dot on Ing ayl); and the lexicalized pa/7/ s
pa'ta'ng ay (1462 'Nung 10:79a), paflj spa'tang (1527 Cahoy l:15a=29a).
pal s kalak ‘toe’: pal s kala k ol (1482 Nam 1:50a); and the lexicalized pa/7/ s kala k o 'Iwo (1447
Sek 6:39a ['Yi Tonglim version]).
Ramsey concludes that all cases of High-High or High-Low belong to a single class of words with
optional retreat (or spread?) of the high pitch, whether both variants are attested or not. In this book, a
noun attested in only one variant is cited in that form; those with two variants are cited as one or the
other, depending on the distribution. In effect, we imply that the earlier language had two classes,
which eventually fell together, as indicated by the accent classes of the modern dialects described by
Ramsey. To the extent we differ with Ramsey, it is perhaps a question of the timing of the merger of
patterns, but we end up with at least four accent classes and he has only three: Low-Low (as in toll
‘bridge’, mozom ‘mind’, motoy ‘joint’, polom ‘wind’, pwuthye ‘Buddha’, ... ), Low-High (as in kwo'ki
‘fish, meat’, na'lah ‘nation’, se'li ‘frost’, a'tol ‘son’, a'chom ‘morning’, ... ), and High with the pitch
of the second syllable nondistinctive as in these examples:
HH / HL
‘child’ 'a'ki (1447 Sek 9:25b; 1459 Wei 1:44b, 8:100b, 8:101b, 21:124b; 1463 Pep 6:47a; ’1517- Pak
1:56a, 1:57a, 1:57b), a'kfi] oy (1459 Wei 8:81b, 8:83a); 'aki (1447 Sek 6:13b; 1449 Kok 148;
1459 Wei 2:33b, 8:86a, 8:86b, 10:24b, 23:74b, 23:87a; 1475 Nay 2:1:40b; 1485 Kwan 10b)
‘drought’ ko'mol (1527 Cahoy l:3a=2a), ko'mol s (1447 Sek 9:33b, 1462 'Nung 8:115a); komo'l
ay (1445 'Yong 2, 1463 Pep 2:28a), 'komo'l i (1481 Twusi 7:36b, 25:11b)
‘firefly’ 'pantwoy (1446 Hwun 29ab; 1465 Wen 1:2:3:40b, 2:2:1:52a; 1481 Twusi 8:40a, 21:9a,
24:7a; 1482 Nam 2:59b); 'pantwoy lol (1481 Twusi 6:20b)
= 'pan two (1527 Cahoy l:llb=21b); 'pantwoy ’lwo[']ta or 'pan'two ’ylwof'Jta (1481 Twusi
8:12b)
‘food’ ppen] >) 'cha'pan (1447 Sek 6:16a, 1527 Cahoy 2:10a=20a), 'cha'pan ol (1449 Kok
122), 'cha'pan 'two (’1517- Pak 1:7a); 'chapan (1459 Wei 1:32a), 'chapa'n on (1459 Wei 2:25b),
chapan ul (1481 Twusi 24:63a); [Jchapa'n i (1481 Twusi 22:6a)
‘granny’ 'hal-’mi (1459 Wei 10:17b, 1527 Cahoy l:16a = 31a), 'hal-’mi lol (1445 'Yong 19); hal-
’m[i] oy (1482 Kum-sam 3:12a; 1482 Nam 1:8b, 2:4a)
‘lightning’ 'pen'key (1445 'Yong 30, 1447 Sek 6:32a, 1449 Kok 161, 1463 Pep 3:35a, 1482 Kum-sam
2:44a, 1527 Cahoy 1: lb=2b); 'penkey (1465 Wen 2:1:2:19a), 'penkey s (1482 Kum-sam 2:44b)
A Reference Grammar of Korean
PART I 67
‘mark’ 'pwo'lam (1463 Pep 5:14a, 1527 Cahoy l:18b = 35a), pwo'lam i ( ? 1517- 'No 2:16a),
pwola'm ol ( ? 1468“ Mong 20a); 'pwolam (~ ho'ya 1462 'Nung 1:70b, ~ ho.ya 1482 Nam 1:
70b), pwo'la.m in i (1459 Wei 21:217-8, 1462 'Nung 8:119b)
‘mother’ 'e'mi (1527 Cahoy l:16a=31a; 1459 Wei 8:86a, 21:22a, 21:27a, 21:53a; 1482 Kum-sam
2:61a; 1481 Twusi 8:67b), e'mi lul (1518 Sohak-cho 9:55a), e’mi /’ (1462 'Nung 5:85b), emft]
uy (1459 Wei 21:21b), fjemi i'sya (1459 Wei 8:83a); 'emi 'lol (1447 Sek 6:1b; 1459 Wei 21:20a,
93a), emi two (1447 Sek 6:3b), emi 'Gwa (1462'Nung 5:85b); 'emi ’ la (1459 Wei se: 14a),
‘net’ ku'mul ( ? 1517- Pak 1:70b; 1527 Cahoy 2:17a), ku'mu.l i'la (1464 Amitha 7a); kumu'l ey
(1447 Sek 9:8a, 1462 'Nung 8:93a), 'kumu'l i (1459 Wei 8:10b), kumu'l un (1481 Twusi 7:3a),
kumu 'l ul (1463 Pep 2:8b, 2:24b), kumul s (1481 Twusi 21:38a)
‘rabbit’ thwos'ki (1465 Wen 1:68a, 1527 Cahoy 1:10b = 19b), thwos'ki /’ (1481 Twusi 16:36b);
thwos.ki (1481 Twusi 21:38a, 1482 Kum-sam 4:63a), thwos.ki lol (1481 Twusi 10:26a),
thwos.k[i] uy (1462 'Nung 1:74a, 1466 Kup 1:6a, 1482 Kum-sam 2:66b) = thwos.kfi] oy (1481
Twusi 24:25b, 1482 Kum-sam 4:36b)
‘shade’ konol (1462 'Nung 8:50b, 1463 Pep 6:165a, 1527 Cahoy 1:1a). ko'nol'h i (1462 'Nung
8:5la)l ko'nol'h ay (1465 Wen 3:1:2:50a); konol i (1463 Pep 5:180a), konol Gwos (1481 Twusi
23:8a), konol'h ay (1463 Pep 2:103-4, 1481 Twusi 7:24a), konol'h i (1459 Wei 18:26a, 1463 Pep
3:45b, 1481 Twusi 24:30a), konol.h on (1481 Twusi 15:9b)
‘sleeve’ (1) 'so'may (1527 Cahoy 2:1 lb=23a); 'somay (1481 Twusi 8:45b), somay lol (1481 Twusi
20:47a, 22:25a), somay 'yey (1481 Twusi 6:4a, 23:2a),
(2) 'so'moy 'yey ( ? 1517- Pak 1:72a), so'moy s (1463 Pep 1:31b)
‘snake’ poy'yam (1527 Cahoy l:lb=22a), 'poy'ya.m i'Gwo (1459 Wei 21:42b); poyya'm i (1445
'Yong 7), 'poyya'm oy (1463 Pep 2:165b), 'poyya'm on (1481'Twusi 21:38b), poyya'm ol (1481
Twusi 15:8b), 'poyyam 'kwa y (1462 'Nung 7:79a, 1481 Twusi 6:4a), poyyam 'kwa (1550
'Yenghem 15b)
‘thunder’ 'wul'Gey (1447 Sek 6:32a, 1463 Pep 3:35a, 1482 Kum-sam 2:2b, 1527 Cahoy 1:2b),
'wul'Gey s (1481 Twusi 10:18a); wulGey (1481 Twusi 7:24b, 10:19a; 1482 Nam 34b)
‘topknot’ 'syang'thwu (1527 Cahoy 2:12b=25a) = syangthwo (1586 Sohak 2:2a)
‘twenty’ su'mul sal (1445 'Yong 32) ‘20 arrows’, su'mul kwo't ile'la (1447 Sek 6:38a) ‘it was 20
places’, su'mulfli] hon hoy s so'zi yey (1447 Sek 6:47a) ‘in the space of 21 years’; su'mul.h i'm
ye (1462 'Nung 2:57b), sumul'h in ssi-'cyelq 'ey (1462 'Nung 2:8b) ‘when 20 (years old)’, 'su'mu
nal ^ 1517- Pak 1:8b) ‘20th day’, su'mu na'mon hoy lol (1447 Sek 24:2a) ‘for over 20 years’,
su'mu "nas ( ? 1517“ Pak 1:20a), su'mu lyang 'two (id. 1:20b); su mul'h ey sye (1462 'Nung
2:6b), but sumul'h ey (ibid., also 1481 Twusi 8:19a)
‘wolf’ ilhi (1527 Cahoy 3:10a = 18b); ilhi Gwa (1447 Sek 9:24b) ‘and wolves’; ilhi towoy'ye
’ys.two'ta (1481 Twusi 10:19b) ‘has become a wolf’.
Not in Ramsey’s list of HH / HL nouns:
‘bowl’ swo 'la s ( ? 1517- Pak 1:56a); swola lol (1586 Sohak 2:3b)
‘mosquito’ 'mwo'koy (1447 Sek 9:9b, 1459 Wei 9:26a), mwo'kuy (1527 Cahoy 1:1:1 lb=22a);
'mwokoy swo li ? < 'mwokoy fsj swo li (1462 'Nung 4:3b); f 'Jmwo 'koy (1579 Kwikam 2:60a)
‘net-edge guide ropes’ pye'li (1527 Cahoy 2:8a= 14b); 'pyeli lol (1481 Twusi 16:63b)
‘now’ icey [< i cek ‘this time’] (1447 Sek 6:5b, 6:1 lab, 24:16a; 1459 Wei se:13b, 2:42b, 2:64a,
8:98a, 8:101a, 9:35f, 10:8b, 13:19ab, 21:21b; 1481 Twusi 7:31b, 8:10b, 8:38a; 1482 Nam 1:30b),
icey n’ (1463 Pep 2:41a), 'icey s (1447 Sek se:6b, 1459 Wei 2:9b, 9:35d), 'icey 'two (1459 Wei
2:64a), 'icey 'za (1449 Kok 115), 'icey ’ la (1459 Wei se: 13b); i cey (1462 'Nung 4:126b, ? 1517-
'No 1:1a), i cey /’ (1462 'Nung 10:19a). Also icey (1459 Wei 23:78a), icey n’ (1463 Pep 5:178b).
‘pillow’ 'pye'kay (1527 Cahoy 2:6b = 1 lb); 'pyekay (1463 Pep 2:73a, 1481 Twusi 15:11b)
‘picture’ ku 'lim (1527 Cahoy 3:9v=20v); kulim (1481 Twusi 16:25b, 7 1517- Pak 1:64b)
Note that kuli'mey (= kulim'cey) is a noun meaning ‘reflection, image, shadow’ and is not to be
taken as ku(')lim ‘picture’ + particle 'ey, for which we lack an example.
68 PART I
A Reference Grammar of Korean
‘plow’ 'ko'lay (1527 Cahoy 1:6a= 1 lr); 'kolay (1481 Twusi 16:39a); kolay (1446 Hwun 28a, 1481
Twusi 25:22a)
The noun 'hyenma ‘how many /much’ has both syllables high in 'hyen'ma s with the genitive particle.
For certain words the accent-marked attestations are really too few for us to draw conclusions:
‘thunder’ 'pye'lak (1527 Cahoy 1: lb=2b) is also attested (says LCT) in 1481 Twusi 18:19, but that
text is not available to me;
‘flute’ phi'li (1527 Cahoy 2:16r=32b) is reported also in 1481 Twusi 9:40, to which I lack access;
‘lotus persimmon’ 'kwo.ywom (1446 Hwun 28b) appears also with the odd pattern HR 'kwo~ywom
(1527 Cahoy 1:6b = 12a) and it is HH 'kwo'ywom in 1517 Saseng 2:13a (says NKW, but the Taycey-
kak repro lacks the dots).
The first-person plural pronoun 'wuli (1447 Sek 6:5a; 1459 Wei 13:35b, 13:36a; ? 1517" Pak
1:54a; ... ) suppresses the second accent when followed by a particle: 'wuli Gwos (1459 Wei 2:72a),
'wuli 'two (1459 Wei 8:100a), 'wuli 'za (1463 Pep 5:121b), wuli n’ (1459 Wei 2:69b), 'wuli 'tol'h i
(1447 Sek 9:40a, 19:30b; 1459 Wei 10:12b, 10:31a, 18:18b), 'wuli 'tol'thwo [= 'tolh 'two] (1459
Wei 18:3a), 'wuli 'uy (1463 Pep 2:231a). Modifying a noun in the sense ‘our’ the word is often 'wuli
(1445 'Yong 3; 1459 Wei 18:42b, 21:193b; 1462 'Nung 10:42b; 1463 Pep 2:5b; 1482 Nam 1:54b;
? 1517- Pak 1:51a) but there are also examples of wu li (1447 Sek 6:5b; 1451 Hwun-en la; 1459 Wei
2:69a, 2:70b) which might be treated as compressions of wuli 'uy. Some cases where a single dot
occurs for an earlier double dot may be due to broken type (as seen from the placement of the
remaining dot), but often these result from historic change, whereby the double-dot (low rising) accent
merged with the simple high accent represented by the single dot, so I have generally left these the way
the text carries them.
The hypothesis that (at least "by a certain time) the high pitch was distinctive only in the first
syllable of a word will account for some of the variant accents found for verb forms in §2.12.4:
ho(')ya ‘do’, ti(')ye ‘fall’, pwuy(')ye ‘cut’, 'psti(')kwo ' insert’, psu(’)ti ‘use’, ptu(')m ye ‘float’.
But the majority of the HH- stems do not exemplify the variation. The hypothesis that there was no
distinction between LHL and LHH could account for accent variants in a few verb forms:
pwo'nayya (1481 Twusi 25:27b) = pwo'nay'ya (1447 Sek 24:15ab) ‘send’
te'pule (1481 Twusi 7:37a, 20:29a) = te'pu'le (1447 Sek 6:23a, 13:15a; 1459 Wei 2:6b) ‘take
along’
i'kuyti (1481 Twusi 8:42a) = i'kuy 'ti (1481 Twusi 7:7b, 1586 Sohak 2:9b) - Cf i'kuy'ye (1459
Wei se:9a) ‘win’
e'wulGwo (~ two 1449 Kok 134) = e'wul'Gwo (1462 'Nung 3:38a) ‘join’
ne'kikwo (1481 Twusi25:23a), ne'kiti (1481 Twusi 16:61b, 1482 Kum-sam 5:14a, 1475 Nay se:6a)
= ne'ki kwo (1462 'Nung 1:34b), ne'ki ti (1447 Sek 24:3ab, 1475 Nay 1:17a) ‘deem’
"te'leWun (1447 Sek 13:33b) = "tele'Wun (1459 Wei 9:24a, 1459 Wei 18:39ab), "tele'wun (1462
'Nung 4:38b) ‘dirty’
il'Gwusyan (1447 Sek se:5b); Cf il'Gwu'sya (1459 Wei 21:218b) ‘achieve’
Peculiarities in the accent patterns of certain pronouns and indeterminates are not amenable to
generalization (Cf Ramsey 1978a: 170-4, 'Yi Sangek 1978:112-6); the attested facts are stated in the
individual entries of Part II.
2.12.3. Accent suppression before particles.
A number of nouns suppress the high pitch on a syllable before the locative-allative particle '%ry
and its variant '%>’ (He Wung 327). Included are most of the monosyllabic nouns that have the high
pitch, and at least one two-syllable noun:
kalh ‘knife’-* kal'h ay (1466 Kup 1:82a)
kilh ‘road’ — kil'h ey (1447 Sek 6:3b, 6:15b; 1481 Twusi 7:6a, 8:2b, 10:27b; 1482 Nam 1:49b;
? 1517- Pak 1:54a)
kwoh ‘nose’ -* kwo'h ay (1447 Sek 13:38b, 1459 Wei 1:36b), kwo'h ay 'sye (1462 'Nung 3:24b)
'kwuy ‘ear’ -*• kwuy 'yey (1447 Sek 19:16a, 1449 Kok 2, 1475 Nay 1:37a)
A Reference Grammar of Korean
PART I 69
mwom ‘body’ -*• mwo'm ay (1447 Sek 19:19b, 1462 'Nung 10:18a, 1481 Twusi 8:33-4) and
mwo'm ay s (1447 Sek 9:12a, 1459 Wei 2:53b)
moyh ‘moor’ -* moy'h ay (1463 Pep 6:154b, 1481 Twusi 7:30a), moy'h ay s ( ? 1468- Mong 27;
1482 Nam 1:4a, 1:49b; 1482 Kum-sam 3:34b), moy 'h ay sye (1481 Twusi 7:39a)
nac ‘daytime, noon’ -> mac oy (1445 'Yong 101)
'nwun ‘eye’ -*• nwu'n ey (1481 Twusi 25:9b, 25:47a)
pich ‘light’ -* pi ck ey s (1447 Sek 19:18a ), pi'ch ey sye (1481 Twusi 8:9b)
ptut ‘meaning, mind, intention’ -*• ptu't ey n’ (1447 Sek 19:34a ),ptu't ey s (1447 Sek 9:26b)
tet ‘time’ -*• a'ni han tet ey (1463 Pep 4:32a, ? 1468~ Mong 26b, 1485 Kwan 9a) ‘in a short while’
pam ‘evening, night’ -*■ pa'm oy (1447 Sek 6:19b, 1462 'Nung 1:16b), pa'm oy two (1459 Wei
2:27a), pa'm oy f'Jza (1481 Twusi 23:6b)
pwom ‘spring’ -» pwo[']m oy (1482 Kum-sam 2:6b; the repro obscures the dot)
'stah -+ sta'h ay (1449 Kok 41; 1459 Wei 1:28b; 1462 'Nung 8:123a; 1481 Twusi 7:7b, 15:45a,
21:42b, 25:43a; ?1517- Pak 1:64b)
swo'li ‘sound’ -»• swoli 'yey (1481 Twusi 7:39a)
For at least one word this holds for the genitive uses of the particle '%y, too: mom ‘another person’
-»■ no'm oy + noun phrase (1447 Sek se:6a, 1463 Pep 2:28b, ? 1517- Pak 1:9b, 1465 Wen 3:3:1:62a,
? 1468“ Mong 20b, 1518 Sohak-cho 8:15a [no'm is smudged], 1475 Nay 1:9a); and also as the genitive-
marked subject of an adnominalized verb (1465 Wen se:77a). That differs from nwom ‘lowly person’
where the genitive is nwo'm oy (1459 Wei 17:76b), later reduced to nwo.m oy (1481 Twusi 7:6b).
Monosyllabic nouns which do not lose their accent before the locative: 'hye ‘tongue’, hoy ‘sun’,
'poy ‘belly’, pi ‘rain’, mul ‘water’, mwul ‘crowd’, pul ‘fire’, pol ‘community, village’ as in
'i 'po l ay ‘in this village’ (1459 Wei 8:94a), skwum ‘dream’. Kim Wancin would assign these
exceptions an underlying pattern of High-High, rather than just the single High assigned to the other
accented monosyllables. Perhaps the high persists from a lost or absorbed second syllable?
The word for ‘bosom’ is not attested without the locative particle, but we infer that phwu m ey is
from *'phwum on the basis of the attested accent of the related verb phwum- ‘embrace’.
In the same environment (before the locative marker) the double dot is sometimes reduced to a
single dot, i.e. the low-high rise becomes just high:
"mwoyh ‘mountain’ -* 'mwo.h ay s (1482 Kum-sam 3:36b) = "mwoy'h ay s (1482 Kum-sam
3:33a); but "mwoy'h ay (1449 Kok 41), "mwoy'h ay s (1481 Twusi 7:30b)
"swok ‘deep inside’ -* swo.kay s (1481 Twusi 7:24b) but "swo'k ay (1459 Wei 1:13a)
"twolh ‘beam’ -* 'twol.h ay (1481 Twusi 7:5a) ‘to the beams’
Usually the double dot is retained: "nwu'n ey s tol (1482 Kum-sam 2:61b) ‘moonlight on the
snow’; 'mul s "ko'z ay (1459 Wei 8:99a) ‘at the water’s edge’; (••• s) "i'l ey (1475 Nay 2:2:47b) ‘in the
event (of - )’; "twuy'h ey 'sye (1445 'Yong 28), "twuy'h ey n’ (id. 70), "twuy.h ey nun (id. 30),
"twuy 'h ey (s - ) (1459 Wei se:24b) ‘in back’.
The modern dialects of Hamkyeng and Kyengsang show a similar cleavage of monosyllabic tonic
nouns, and the grouping is probably inherited from the 15th-century accent, but we lack sufficient data
to set up a system of correlations that will account for the exceptions. (See Ramsey 1978a: 167-9.)
The accent of a monosyllabic noun is sometimes lost before the genitive particle s, perhaps
evidence that certain cases of Nj s N 2 are compound nouns: "mwoyh -* mwoy[h] s "kwo'l ay (1447
Sek 6:4b, 1449 Kok 141) ‘in a mountain valley’.
2.12.4. The accentuation of verb forms.
There are many complexities in the accentuation of MK verb forms and the corresponding forms
in the modern dialects. Studies by He Wung, Kim Wancin, Ceng Yenchan, Kim Chakyun, 'Yi Sangek,
and others shed light on many of the problems, and in particular Ramsey 1978a presents a wide view of
the situation and discusses the interpretation of the available data in admirable detail.
We must assume a basic accent for the endings -' u /om, - ’%/<?, and - u 4m so as to account for such
phrases as pa'to'm ye (1462 'Nung 8:104b), me'kulq tet (1459 Wei 8:8b), and two'to'n i (1445 'Yong
70 PART I
A Reference Grammar of Korean
101) even though the accent is often or always suppressed in many structures. That is why there are
discrepancies between the accent of entry citations and examples for some of the forms in Part II.
Stems fall into two major types, depending on whether they begin high (single dot) or low (no dot
or double-dot), but many of the monosyllabic low-accent stems that do not end in a consonant take on
high pitch before certain endings - or, put another way, many of the monosyllabic high-accent stems
that end in a vowel lose the accent in many of the common paradigmatic forms, such as the gerund
-'kwo, the adverbative - key, the indicative assertive -'ta, the suspective - ti, the hortative - eye, the
substantive which appears only in the complex structure -('%)'m ye, the modifier -('%)« and
prospective modifier -('%)/(#) and forms based on these (including the subjunctive attentive -la, as in
ho'la ‘do it’). But they retain the accent before the infinitive the honorific -(%) si-, the
deferential -"zop-, and the aspect markers -(’)no-, - 't e/ a-, -'k £/ a-. This seems to indicate that the
infinitive, like the other markers, was originally a bound stem. (And that deepens my suspicion that
is cognate with Japanese a[rj- ‘be’.) The polite marker -ngi is like the bound stems, to judge from
(■■■'%.) 'ci-ngi ’’ta ‘wants (to do)’ < °ci-, i-ngi ’’ta (1459 Wei 21:218b) ‘it is ’ < i-, and kwo'c
i phu-'tos ho-ngi ’’ta (1463 Pep 1:85b) ‘the flowers seem in bloom’ - but also "ep-'tos ho-ngi ’’ta
(1462 'Nung 1:105b) ‘seems to lack’. If so, the lack of an initial dot on ha-ngi ’’ta (1463 Pep 7:68b,
1464 Kumkang 62b) ‘are many’ < °ha- is puzzling. The high/low stems are basically low before the
modulator (-'u'%- etc.), though after most vowels that is obscured by the usual ellipsis of that
morpheme, which leaves behind an accentual trace:
"na'toy (1447 Sek 19:7b, 1449 Kok 185) *- na-'[wo]-'toy, "nalq (1462 'Nung 3:24b) *- na-'[wo]lq
< °na- ‘emerge’
"ha m ol (1482 Kum-sam 3:19a) *- ha-'[wo]-m < °ha- ‘much/many’
"cwulq (1447 Sek 9:12a) *- cwu-'[wu]-lq < °cwu- ‘give’
"wo'm i (1459 Wei 9:10b, 1482 Kum-sam 3:19a), "wo'm ol (1482 Nam 1:50b), "wo[']m ay (1481
Twusi 21:25b) *- wo-'[wo]-m < °wo- ‘come’
"pwom (1462 'Nung 2:84a) *- pwo-'[wo]-m < °pwo- ‘see’
But after the modulator survives intact:
ni'ywu'n i (1481 Twusi 7:1a), ni'ywun (id. 10:18b) *- ni-'wu-n, "niywon (1482 Nam 1:72b; ?=
ni'ywon ) *- ni-'wo-n < °ni- ‘roof, thatch (a roof)’
"cywu'm un (1462 'Nung 8:8a) = Q*)ciywu‘m un *- ci-'wu-m < °ci- ‘carry on the back’
"cywu'm ey (1481 Twusi 7:6b) = C*)ciywu'm ey *- ci-'wu-m < °ci- ‘cut (wood)’
There is something odd about the accent of 'iywo'm on (1482 Nam 2:64a) where we expect *i'ywo'm
on *- i-'wo-m < °i- ‘carry on the head’; perhaps the scribe misplaced the dot.
Compare the modulated forms of those stems that are always high:
'thywo'toy ( ? 1468“ Mong 53a) ■*- 'thi-'wo-'toy, 'thywon (id. 10a) «- 'thi-'wo-n, 'thywum (1463 Pep
5:38a) *- thi- 'wu-m < 'thi- ‘hit’
'skoy'ywom (1462 'Nung 10:1b) *- 'skoy-'wo-m < skoy- ‘wake up’
Before the causative formative - i- the high/low stems are basically low, and that accounts for the rising
accent of some of the stems of group la below: "nay- ‘make emerge’ < *na-'i- < °na- ‘emerge’,
"pwoy- ‘show’ < *pwo-'i- < °pwo- ‘see’, "syey- ‘erect, let/make stand’ < *sye-'i- < °sye- ‘stand’,
"tiy- 1 ‘drop, let/make fall’ < *ti- i- < *°ti-\ ‘drop’, and "tiy-1 ‘smelt (metal), create (out of metal)’
probably < *ti-'i- < °ti~2 ‘become’. The basic final y on "tiy-\ ‘drop’ and "tiy-2 ‘smelt, create’ is
needed to account for the velar lenition in such forms as "ti 'Gwo\ (1459 Wei 10:24b) ‘dropping’ and
the unattested *"ti'Gwo2 ‘smelting, creating’. (The only other case of basic iy- is 'iy-, the copula, with
the gerund form 'i'Gwo.) The summative -ki is nonleniting, and that accounts for 'swoy ku'lus "tiki
yey s 'swo.h i'la (1465 Wen 1:1:2:181a) ‘it is a mold for making metal vessels’.
Interestingly, when the summative - ki started taking over part of the work of the suspective - ti, it
was treated not like - ti or - key, but like the bound stems: ka'ki ( ? 1517- 'No 1:26b), 'pwo'ki (id.
1:37b) - compare ka'ti (id. 2:7a), pwo'ti (below). The earliest examples of the summative, however,
are what we expect: ho'ki lol cul'kye (1447 Sek 6:13a), ho'ki lol culki'ti (1459 Wei 10:18b). And
A Reference Grammar of Korean
PART I 71
that 16th-century text also has ho'ki Gwa ( ? 1517- 'No 2:43b), so the first two examples above may be
scribal errors. Another regular example: "il ho 'ki yey (1481 Twusi 25:7b).
The transferentive, to my surprise, is treated as if a bound stem: kata ka (1445 'Yong 25, 1482
Nam 1:36b), wota 'ka (1447 Sek 23:57b), 'hyeta ka (1481 Twusi 16:1b), hota ka (1462 'Nung
3:84a, 1482 Kum-sam 2:31b), na'ta ka ’m ye (1459 Wei 21:215b), 'tita 'ka ’m ye (1481 Twusi
25:43a).That argues in favor of the notion that -taka is a bound infinitive rather than the
indicative assertive -'ta + particle 'ka, as it is viewed in this book. The only apparent counterexample,
in na-ka'kwo eye tha ka ••• (1481 Twusi 8:29a) ‘I want to go out but - ’, is probably a surface
reduction of 'na-ka'kwo 'eye hota ka. Also arguing for explaining the transferentive as a bound
infinitive is the accentuation of the low-pitched stems that are closed monosyllables:
makta ka k (1466 Kup 2:66a), and not *mak'ta ka k
cwukta ka (1445 'Yong 25) despite cwuk'ta (1459 Wei 17:21a)
is.ta ka (1482 Kum-sam 2:13b), and not *is'ta 'ka despite is'ta (1462 'Nung 2:83a;
1463/4 Yeng 2:62a; 1482 Kum-sam 3:9b; 1482 Nam 1:14a; ?1517- 'No 1:62b, 2:36a)
There is another set of structures on which the accent sheds light. What I had earlier taken as the
indicative assertive - 'ta (and i'la ) + forms of the emotive bound verb sfoj- I now realize must be
retrospective emotive forms: -'taswo'la, -'taswo'n i, -'ta'songi ’’ta, -u'l i ’'la[']s-ongi ’’ta, -u'l i
’laf'Jsongi ’’ta. (See the entries of Part II, where they are so treated.) That is because of the accent of
the stem in such forms as hotaswo'la (1481 Twusi 16:18a), ho'taswo'n i (1446 Sek 13:43b), and
ho ta's-ongi ’’ta (1462 'Nung 2:6-7).
In citing a stem or the “naming” form (the indicative assertive) of the high/low verbs we will use
a hollow dot (°), meant to represent zero (the low pitch) except when one of the relevant elements is
attached — the infinitive ending or one of the bound stems. With the prominent exception of °ho- ‘do;
be’ there are no high/low stems ending with the minimal vowel %, for those are all high; the stems
that end -e- or -a- are all of the high/low type. (There are no low-pitch monosyllabic stems ending
- %-, but there are such stems with the shapes and ”-%y-.) This fact may be used to argue
that the vowel of °ho- must be a reduced form of some other vowel. Elsewhere I use the irregular
infinitive to support the claim that the stem was earlier the unique shape *hyo- (see the entry °ho ta in
Part II). It may be questioned whether the semivowel adequately strengthens the minimal vowel, but
compare the monosyllabic stems that end in which all belong to the high/low group with the
unexplained exception of 'skwu- ‘dream’, which may well be contracted from a dissyllabic stem (as
suggested by the initial cluster) and 'hwo- ‘broad-stitch’, for which the modern ho- (and dialect accent
corresponding to the long vowel) would suggest an earlier version *°hwo- like °pwo- ‘see’.
We find the following groups of monosyllabic stems that do not end in a consonant:
(1) Stems that are rising, here marked with a preceding high double dot ("•—). None end in -%y-
(without preceding w) or in nor in -vv%-; the apparent exceptions such as "cwu- or "pwo- are
modulated forms of non-rising stems (°cwu-, °pwo-).
(la) Rising in all forms. These stems end in -ay-; -ey-, -yey-; -w u /oy-. Also "hoy-, the causative
of °ho- ‘do’, and two stems ending in basic -iy- that were originally causatives, too: "tiy-\ ‘drop,
let /make fall’ and "tiy- 2 ‘smelt, create (out of metal)’. The modulated versions of simple °-a-, °-e-,
°-wo-, and °-wu- also belong here: "ka- *- ka-'[wo]- < °ka- ‘go’, "sye- *- sye- [wu]- < °sye-
‘stand’, "pwo- *- pwo-'[wo]- < °pwo- ‘see’, "cwu- *- cwu-'[wu]- < °cwu- ‘give’.
(lb) Rising in most forms (including the effective forms with -G e ki-), but not the infinitive and
forms containing the modulator or the short version of the effective aspect marker - '(y)^h-. These stems
end in -w u /oy-, -ay-, -ey-, -yey-. We mark them with two hollow dots (’’--).
(2) Stems that are always low, here left unmarked ( —). These stems end in -%y- or •••w%y-.
(3) Stems that are always high, here marked with a preceding high dot ('•—). These stems end in -i-,
-w%y-, -oy-, or (?) -uy-; also the copula 'iy-. The stems 'skwu- ‘dream’ and hwo- ‘broad-stitch’
exceptionally belong here, rather than in group 4.
(4) Stems that are high/low, here marked with a preceding hollow dot (°—). They are low except in
72 PART I
A Reference Grammar of Korean
the infinitive and when compounded with other stems, including bound stems such as the honorific
(- %'si-), the aspect markers (-'k^a-, - 't e h-, -'no -), the deferential (-" zop-), and the polite (-ngi
These stems end in -a-, 1-e-, -ye-, -w u 4>, (Apparently there are no monosyllabic stems that end
in -e- without a y before the vowel.) The stems stay low when - 'two ‘even/too’ or -'tos ‘like’ is
attached.
The accent groups to which a stem of a given shape may belong:
la "...-
lb “...-
2 ...-
3 '...-
4
■i-
"tiy- 1,2
+
+
■ %-
+
°ho-
■ %y-
"hoy-
+
+
■wu-
0
'skwu-
+
■wo-
0
"hwo-
+
■w'Mjy-
+
+
+
+
■ye-
0
+
■ey-
+
+
+
■yey-
+
+
■a-
0
+
■ay-
+
+
+
RISING
RISING/LOW
LOW
HIGH
HIGH/LOW
The parenthesized blanks are modulated stems (stem + modulator) only.
There are stems that end in ™i- both in Group 3 and in Group 4. Only the effective forms of °ni-
‘go’ are attested: hike (1445 'Yong 58, 1459 Wei 8:1a), hike (1459 Wei 8:101b); hike'la (1459 Wei
8:101a); hi'kesih i (1459 Wei 8:93ab); hi'kesi'tun (1445 'Yong 38); hikehol (1463 Pep 4:37b).
From these hi-ke- forms alone we cannot tell whether the stem belongs with Group 3 (HIGH) or group
4 (high/low), but we assign it to Group 4 to accord with °nye- ‘go’, from which it was likely derived.
The somewhat later stem "nyey- seems to belong with Group la (always rising).
A number of accentually anomalous examples have to be explained individually, as prosodic
adjustments or scribal mistakes: 'wuy []ho'sya (1463 Pep 7:17a); i la [Jhosya'l i (1447 Sek 6:17a),
kot f'Jhosi'n i (1463 Pep 2:43b); hoh i [JGwo (1481 Twusi 7:40a), "mwot ho'si.l i ’ la (1462 'Nung
2:50b); "ep-'tos [Jho-ngi ’’ta (1462 'Nung 1:105b); solang f'Jhosi'nwon pa lol (1475 Nay 1:55b);
"mwot ho’ya 'ys.ke'nul (1462 'Nung 8:57a); a'ni ho.ya is'ta.n i (1481 Twusi 7:23a), a'ni ho'ya
'ys.ta'n i (1465 Wen 1:1:1:44b), a'ni hotwo'ta (1481 Twusi 8:2a); 'QHQ- ttyeng ho 'sa-ngi ’’ta (1459
Wei 8:96b). We would expect the contractions of °ho-k e h- to be kh%.-~ and that is what we find in
'kh^an ma'lon, khesin, khesina, and khesin ma'lon. But most of the other forms are attested only
without the accent: khe'n i ('Gwa, ’ la), khe'nol, khe'ml Cza), khen 't i, (a'ni) khan t i ’.n i ’’la,
khen tyeng (Cf hoken tyeng), khe'ta, khe'tun. And only khe[']n ywo despite kha.n ywo (1481 Twusi
16:37b), khe'n i Gwa despite a'ni 'kha.n i Gwa (1481 Twusi 16:61b). These anomalies are probably
the result of secondary loss of the accent, though the details are unclear. That must be the case, too,
for khe za (1475 Nay 2:1:16b) < hoke 'za; compare 'khe 'za (1463 Pep 2:224b). There are a few
similar cases for the- < hote- , such as ku'le the n i a'ni the'n i (1459 Wei 9:36d) < hote'n i,
a'ni tha n i (1463 Pep 2:28b) < hota'n i, a'ni thwo'ta (1481 Twusi 8:2a, 16:22b) < 'hotwo'ta.
There are two ways to look at the stems of group 4. The usual assumption (He Wung, Kim
Wancin, Ramsey 1978a) says that the stems of the first group are historically low and acquired an
accent before the infinitive, the bound stems, and so forth. ‘Yi Sangek 1978:119 (and now Ramsey
1992) would treat the stems as high, especially because as the first element in compound verbs they are
high regardless of the following stem. (But the form in the compound is often the infinitive.) There are
arguments both ways. Suppose we say that all the stems in groups 3 and 4 were basically high, but that
those with the sturdier vowels suppressed the accent in the paradigmatic forms mentioned. We would
then have to explain why there are examples of -i- in both groups, and there seems to be nothing else
that differentiates these two sets of stems. Several causative and passive stems are derived from
A Reference Grammar of Korean
PART I 73
monosyllabic vowel stems and the derived stems almost all start low, even those from the always-high
stems: pto'i- (1481 Twusi 16:71a) < pto- ‘pick, pluck’, thoy'Gwo- (1459 Wei 7:52b) < tho-
‘receive, undergo’ or ‘ride’, skoy'Gwo- (1481 Twusi 15:26a) < skoy- ‘awaken’ [the first dot of the
LCT entry “ 'skoy'Gwo'ta" is an error], pso'i- (1459 Wei 14:7b) < pso- ‘wrap’, pco'i- (1475 Nay
2:2:51b) < 'pco- ‘weave’, psu'i- (1481 Twusi 23:38a) < psu- ‘use’, su'i- (1481 Twusi 10:39b) <
su- ‘write’, ptuy'Gwo- (1459 Wei 8:99a) / ptuy'Gwu- (1459 Wei 18:56b, 1481 Twusi 22:39b) < ptu-
‘float’, phwuy'Gwu- (1462 'Nung 7:16b, ...) < 'phwuy- ‘burn (a fire)’. The one exception (LCT 693b)
is from a passage poorly reproduced: [?']choy'i- (1466 Kup 2:18a) < cho- ‘kick’. Examples derived
from the high/low stems of group 4: "hoy- < ho i- < °ho- ‘do’, "cay- < *ca'i- < °ca- ‘sleep’, "nay-
< *na'i- < °na- ‘emerge’, sye'i- (1518 Sohak-cho 9:19b) [LCT "syei'ta is incorrect] = "syey- (1445
'Yong 11, 1481 Twusi 15:29b [under wrong entry in LCT]), "pwoy- < *pwo'i- < °pwo- ‘see’. Yet
cii- (1459 Wei 21:106a; Cf LCT 683b) < °d- ‘carry on the back’ is inexplicably high.
The entries in the dictionaries are unreliable guides for many of the verbs discussed here: NKW
has "sye'ta ‘stand’ with an initial dot while LCT lacks the dot; LCT has a dot on °hhye'ta but not on
°hye'ta ‘pull’ while NKW omits the dot for both; neither dictionary has a dot on °sa'ta ‘buy’. Of the
high group, 'hoy'ta ‘white’ has the dot in NKW but not in LCT.
Further complications of stem behavior are largely the result of compressing syllables. They are
taken up in the description of verb conjugations.
2.12.4.1. Vowel-final monosyllabic stems that are rising.
Groug la. Rising in all forms
"pwoy- ‘show’: "pwoyGwo 'eye (1465 Wen se:43b), "pwoyGwo 'za (1447 Sek 6:34ab); "pwoy'm
ye (1462 'Nung 6:89a); "pwoy'ya (1459 Wei 13:35b, 1463 Pep 4:63a); "pwoyye'ton (1462 'Nung
2:23a); "pwoy'Gesi'nol (1449 Kok 110); "pwoyno'ta (1465 Wen 1:1:2:107a); "pwoy'sya (1462 'Nung
2:17a); "pwoyywo'm i (1459 Wei 10:7b), "pwoy'ywo.l i ’ la (1459 Wei 21:21b)
"pewoy- ‘shine’: "pcwoyGwo (1481 Twusi 8:47a); "pcwoyl (1463 Pep 3:12b, 1462 'Nung 4:72b,
1527 Cahoy 3:6a=13a); "pewoy'm ye n’ (1466 Kup 2:14b); "pewoy 'n i (1459 Wei 2:51a); "pewoy 'ya
(1459 Wei 1:48b, 1462 'Nung 3:76a); "pewoynon (1482 Kum-sam 3:59a); "pcwoyywo'm ol (1481
Twusi 7:13b), "pcwoyywo'n i (1481 Twusi 10:31-2)
"nay- ‘make emerge, ... ’: "nay'ti (1463 Pep 2:249b), "nayti ( ? 1468‘ Mong 18b); "nay'Gwo (1481
Twusi 8:30ab); "nay'Gey (1459 Wei 21:20a, 1462 'Nung 1:29a); "nayl ss oy (1463 Pep 3:180ab),
"nay.l i ’’le.n i ’’la (1464 Kumkang 79b) - nay.I i (1463 Pep 5:196b) = "nay'l i; "nay.n i ’’la
(1447 Sek 24:16b); "nay'm ye ... (1459 Wei 7:48a); ", nay'ya (1447 Sek 6:9b, 1449 Kok 49), "nay'ye
(1518 Sohak-cho 10:34b); "nayya'nol (1482 Kum-sam 4:39a); "naynwon (1447 Sek 9:12a); "nayno'n
i (1459 Wei 1:27b); "nay'ywon (1462 'Nung 3:24b); "nay'sya (1463 Pep 6:97a); "naysil (1447 Sek
24:37ab, 1459 Wei 1:11a), ", naysi'n i ’-ngi ’’ta (1445 'Yong 8); "nay'ywol (1481 Twusi 21:42a),
" nayywo'm i (1482 Kum-sam 4:39a), " nayywo.m i'la (1462 'Nung 4:27b)
"kay- ‘get clear’: "kay'Gey (1459 Wei 10:88a); "kayl (1527 Cahoy 3:lb=2a); "kayn (1462 'Nung
10:1b), ["}kayf'Jn i (1481 Twusi 23:20a); "kayGe'nol (1481 Twusi 16:65a); "kaytwo'ta (1481 Twusi
7:7b)
"cay- ‘put to sleep’: "cay'key ( ? 1517- 'No 1:47b; key = Gey)', "cayte'n i (1447 Sek 6:16a). The
infinitive should be * "cay y e/ a, the effective forms *"cay'G e/ a-, and the modulated stem *"cay'yw u /o-.
"pskey- ‘pierce’: "pskey'Gwo (1459 Wei 1:2a, 2:48b); "pskey m ye (1459 Wei 8:24b, 1462 'Nung
1:28a, 1465 Wen 1:1:2:16b); "pskey'n i (1445 'Yong 23, 43), "pskey'n i ’-ngi ’’ta (1445 'Yong 50);
"pskey (1518 Sohak-cho 8:35a, 1463 Pep 5:194b); "pskey'ye (1449 Kok 4), "pskey'ye ti n i (1449
Kok 41), "pskeyye ’ys.ke'tun (1459 Wei 1:27b); "pskeyye'nul (1449 Kok 41); "pskeyywo'm ol (1465
Wen 1:1:1:76a); "pskeyGa'la (1481 Twusi 24:37a); " pskey Ganwos'ta (1481 Twusi 24:26b);
"pskeyzo'Wa (1459 Wei 1:6b); "pskeysi'n i '-ngi’’ta (1449Kokl4); "pskeytwo'ta (1462Kum-sam3:48a)
"mey- ‘shoulder, bear’: "mey'ti (1465 Wen 1:1:1:90a); "mey'Gwo (1465 Wen 1:1:1:90a, 1586
Sohak 6:66a); "meyl (1527 Cahoy 3:10b =23b); "mey[']m ye (1963/4 Yeng 2:73b); "mey'syam
74 PART I
A Reference Grammar of Korean
(1463 Pep 4:79a); "meyte'n i (1449 Kok 119); "mey'zoWa (1459 Wei 10:10b), "mey'zoWa 'za (1459
Wei 10:12b), "mey'zoWo'n i (1459 Wei 10:12b); "mey'zoWwo'l ye (1459 Wei 10:12a),
"meyzo'Wwo.l i ’'la (1459 Wei 10:10b)
"sey- ‘be strong’: "sey'm ye (1459 Wei 1:28a); "seyn (1459 Wei 2:6b); "seyl (1459 Wei 10:30a);
"seysil (1449 Kok 40). This assignment assumes that the unattested infinitive would be * "sey y%, the
effective forms * "sey'Ge- or * "sey 'y C/ a-, the modulated stem *"sey'yw u fo-.
"hyey- ‘reckon, count; think, consider, figure’: "hyey'ti (1459 Wei 17:34b; 1463 Pep 3:62b);
"hyey'Gwo (1459 Wei 2:63b); "hyey'Gwo k (1481 Twusi 15:4a); "hyey'Gey (1459 Wei 1:19a, 1463
Pep 1:26a); "hyeyl (1447 Sek se:lb; 1527 Cahoy 2:lb=2b, 3:9a=21a), "hyey'l i ’le'la (1459 Wei
1:21a); "hyey'ye (1447 Sek 6:6a, 13:26a; 1459 Wei 7:31b; 1462 'Nung 3:76a) - Is hyey'ye 'two
( ? 1517-Pak 1:61b) an error?; "hyey'Gen t ay n’ (1459 Wei 21:104a, 1462'Nung 1:101a), "hyeyf'JGen
tun ( ? 1517- Pak 1:64a), "hyeyGa'l ye two (1459 Wei 21:14a), "hyeyye.l i Ga (1481 Twusi 10:12a);
"hyeym "hyey'non (1459 Wei 9:13b), "hyeynon ta (1447 Sek 6:8a), "hyeynwo'la (1481 Twusi 15:5b)
"hyey'sya (1445 'Yong 104); "hyeyzo'Wol (1447 Sek se:lb); "hyeyywo'toy (1462 ’Nung 4:123b);
"hyey'ywom (1465 Wen 3:3:1:62a), "hyeyywu'm i (1481 Twusi 21:42a); "hyey'ywon (1447 Sek
19:11b), "hyeyywo'n i (1481 Twusi 14:4b)
"syey- ‘make stand; build’: "syey'Gwo (1447 Sek 6:44b); "syey'm ye (1459 Wei 17:37a); "syey'ye
(1447 Sek 9:19b, 1462 ’Nung 4:123b), "syey'ya (1459 Wei 21:213a), "syeyya two (1482 Kum-sam
3:48b); "syeysi n i (1445 'Yong 11); "syeyno'n i (1459 Wei 18:82b); "syeyzo'Wa (1449 Kok 65),
"syey'zoWo'n i (1449 Kok 10), "syeyzo'Wosi'n i (1449 Kok 34), "syeyzop'nwon (1447 Sek 13:14b);
"syey'ywolq (1462 ’Nung 5:8b); "syey'ywo'm i (1462 'Nung 1:19a), "syeyywo'm ol (1475 Nay
2:2:15b), "syey'ywo'm o'lawa (1459 Wei 23:76b); "syeyywo'toy (1459 Wei 17:37b)
"nyey- ‘go’: "nyey.m ye (1481 Twusi 23:19b); "nyeyywo'toy (1482 Kum-sam 5:38a). Some of the
unattested forms must have been *"nyey'ti, * "nyey 'Gwo, * "nyeyn, * "nyeyl(q), * "nyey m ye,
* "nyey y e/ a, * "nyey 'yw u fom.
"kyey- ‘(time) pass, exceed’: "kyeytwo'lwok (1459 Wei 7:9b); "kyeyGe'tun (1459 Wei 7:31b;
broken type looks like “key”). This assignment assumes that the unattested infinitive would be
* "kyey y e/ a, the modulated stem *"kyey'yw u /o-.
"tiy-\ ‘drop, let/make fall’: "ti'Gwo (1459 Wei 10:24b [twice]), "tiGwo (1481 Twusi 15:14ab);
"tiye (1481 Twusi 7:18b); "ti'sya (1449 Kok 45); "tiGe'tun (1481 Twusi 10:32a); "tiywo'm ol (1481
Twusi 8:57b), ti'ywo'm ul ( ? 1517- Pak 1:44b)
"fry -2 ‘smelt (metal); create (out of metal)’: tif'Jm ye (1482 Kum-sam 2:31a); "ti ye (1465 Wen
2:2:2:24b) = "tif'Jye (1482 Kum-sam 2:30a); "ti'zowo'm ay (1463 Pep 1:220a); "ti'ywun (1465 Wen
1:1:2:181a); "tif'Jki 'yey s (1465 Wen 1:1:2:181a)
"hoy- ‘make/let do’: "hoy'Gey (1462 'Nung 3:115b), "hoy'm ye n’ (1449 Kok 99); "hoy l i ’-ngi
s 'kwo (1464 Kumkang 11a); "hoyn (1482 Nam 1:68b); "hoy'ye (1447 Sek 9:21a, 1459 Wei 9:39a,
1482 Nam 2:5a), "hoyye (1481 Twusi 7:16b, 25:37a); "hoy'non (1463 Pep 1:9b); "hoy'sya (1447 Sek
6:7b); "hoysi'mye (1465 Wen 1:2:2:92b)
The stem "ey- ‘turn’ is attested only in the suspective "ey'ti (1518Sohak-cho 8:2b). If the infinitive
was *"ey'y e fa, it belongs to Group la; if it was *ey'y e fa to lb.
Group) lb. Rising in all forms except the infinitive 1 the modulated forms 1 and the short effective forms
°°pwuy- ‘be empty’: "pwuy'm ye (1449 Kok 18) = "pwuy.m ye (1482 Kum-sam 2:54a); "pwuy.l i
'ye (1459 Wei 1:37a); "pwuyn (1481 Twusi 7:4a, 10:32b) II pwuy'ye (1459 Wei 1:48a);
pwuy'ywu'toy (1462 ’Nung 5:59b), pwuy'ywu'm ey (1447 Sek 13:10a)
°°mwuy- ‘move’: "mwuyta (1459 Wei 2:14a); "mwuy'ti (1462 'Nung 3:9b); "mwuyGwo (1481
Twusi 15:52b) = "mwuy'Gwo (1482 Kum-sam 4:39b); "mwuy'm ye (1449 Kok 172, ? 1468" Mong
42b) = "mwuy.m ye (1481 Twusi 7:23b); "mwuyl (1445 'Yong 2; 1459 Wei se:2b, 2:14a); "mwuyn
(1481 Twusi 15:15b, 1482 Kum-sam 2:18a); "mwuyno'n i (1463 Pep 3:35a); "mwuyte'n i (1449 Kok
172), "mwuyte'la (1481 Twusi 8:10a) II mwuy'ye (1459 Wei se:3a, 1462 'Nung 3:117a); mwuy'ywo.m
i'Gwo (1465 Wen 2:3:2:32a), mwuy'ywum (id. 1:1:2:106-7), mwuy'ywu.m ey (1481 Twusi 7:29b)
A Reference Grammar of Korean
PART I 75
°°cwuy- ‘grasp’; "cwuyGwo (1466 Kup 1:15b), "cwuy'lak (1462 'Nung 1:108b, 113a) II cwuy'ye
(1481 Twusi 25:21a); cwuy'ywom (1462 'Nung 1:109b)
°°swuy- ‘rest’: "swuy'ti (1462 'Nung 8:128a); "swuyGwo (1481 Twusi 22:10a, one of the dots is
faint), "swuy'Gwo eye (1463 Pep 3:83a); "swuy'Gey (1463 Pep 2:203a); "swuyl (1459 Wei 1:48a),
"swuy'l i ’Iwo'ta (1481 Twusi 22:33b); "swuyno'n i (1481 Twusi 16:33b). The unattested infinitive
should be *swuy'ye, the modulated stem *swuy'yw u /o-\ swuy'Gwu'm un (1459 Wei 13:18b), though
listed by LCT under "swuy-, ought to be from the causative stem swuy'Gwu- and mol ol
swuy'Gwola (1481 Twusi 21:44b) must be from a variant of that stem, swuy'Gwo-.
°°woy- ‘wrong; left(-hand)’: "woy'ta (1447 Sek 9:14a), "woyta (1445 'Yong 107; 1482 Nam 1:38a,
38b, 39a); "woy'ti (1482 Nam se:2a); woy'Gwo (1481 Twusi 8:10a) - ? repro error for "woy 'Gwo;
"woy'Gey (1462 'Nung 9:77b); "woy.m ye (1482 Nam 1:39a); "woyl (1527 Cahoy 3:12b=29a;
15a= 34b); "woyn (1447 Sek 6:30a, 9:36a; 1482 Kum-sam 2:34a), "woy'n i (1459 Wei 1:42b, 1462
'Nung 2:59a, 1482 Kum-sam 2:3b), "woy.n i (1482 Nam 1:39a) II woy'ywo.m i (1482 Nam 1:39a),
woy'ywo'm ol (1462‘Nung9:83b; 1463Pep 1:6a, 1:10b). The unattested infinitive should be *woy'y^a.
°°mwoy- ‘accompany, escort’ is attested only with -zoW- attached: "mwoyzo'Wa (1446 Sek 23:31b).
°°twoy- ‘hard; severe’: "twoyn ( ? 1517- Pak 1:18a); "twoy'sya (1459 Wei 10:5a), "twoysya'm ay
(1475 Nay 2:2:63b) II twoy'ywola (1475 Nay 2:2:15a). The unattested infinitive should be *twoy'y^a,
the effective forms * "twoy'G e fa- or * "twoy 'y e/ a-. Other unattested forms: *"twoy'ti and * "twoy Gey.
°°tey- ‘burn’: "tey'Gwo (1459 Wei 7:18a); "tey'n i n’ (1466 Kup 2:7b); "teyGe'nul (1475 Nay
se:4a), "teyGe'n ya (1518 Sohak-cho 10:3a) II tey'ye (1466 Kup 1:9b). The unattested modulated
forms would be *tey'yw% ••• , the short effective forms *tey'ye- .
°°pyey- ‘pillow one’s head on’: "pyeyGwo (1466 Kup 1:61b); "pyey'm ye (1459 Wei 1:17a)
II pyey'ye (1481 Twusi 22:19a); pyey'ywun (1459 Wei 1:17b)
*°°myey- ‘get clogged’: myey'ye (1459 Wei 8:84a, 8:98a) ‘get clogged’, myey'ye ’ys.kwo (1481
Twusi 20:33a), myey 'ye ’ys.twota (1481 Twusi 20:35a). Only the infinitive is attested.
°°syey- ‘get white’: "syey'Gwo (1459 Wei 17:47b, 1463 Pep 5:120b); "syey'm ye (1462 'Nung
2:9b), "syeyn (1445 'Yong 19; 1481 Twusi 8:12b, 10:2b, 10:6b, 21:14a), "syey.n i (1481 Twusi
7:12a); "syeyla ’n toy (1481 Twusi 21:42b); "syeyto'lwok (1481 Twusi 16:18a); "syeytwo'ta (1481
Twusi 7:28a) II syey'ywom two (1481 Twusi 15:49b). The unattested infinitive should be *syey'ye; the
effective forms should be *"syey'G e t- or *syey'y e ii- .
°°pay- ‘destroy, exterminate; capsize’: "payta (1481 Twusi 15:34a); "payti (1481 Twusi 22:37b);
"payGwo (1481 Twusi 20:4b); "pay'm ye (1463 Pep 5:43a, 1475 Nay 1:1b) II pay'ya (1462 'Nung
10:92a, 1463 Pep 1:109a, 1482 Nam 2:57b); pay 'yan ma'lon (1445 'Yong 90); pay'ywo.m ol (1481
Twusi 21:36b), pay'ywo.l i ’'la (1459 Wei 7:46b)
°°cay- ‘swift; deft’: "cay'Gwo ( ? 1517- Pak 1:45b); "cayti ( ? 1517- Pak 1:30a, 1527 Cahoy
2:5a=10a), "cay'n i ’ la ( ? 1517- No 1:12b) II cay'ya (1449 Kok 74, 157). The unattested effective
forms should be * "cay 'G e/ a- or *cay 'y e/ a-, the modulated stem *cay yw %- .
°°say- ‘dawn’: "sayl (1462 'Nung 10:45b, 1527 Cahoy 1:1b); "sayno'n i (1481 Twusi 15:46a);
"sayGe'tun (1447 Sek 6:19a); "say'to'lwok (1463/4 Yeng 1:41b) = "say'two'lwok ( ? 1517- Pak
1:21b). The form sa'ya two (1482 Kum-sam 4:52b) is either a mistake for the expected infinitive
say'ya or made on an otherwise unattested variant stem *sai-. The modulated stem: *say 'yw %-.
2.12.4.2. Vowel-final monosyllabic stems that are low.
Grouj) 2. Stems that are low in all forms
chuy- ‘make/let eliminate’: chuy'ywu’l ye (1459 Wei 13:21a), chuy'ywu.l i ’ n i (id. 13:20b, 1463
Pep 2:206a), chuy'ywu.l i ’la (1465 Wen se:47a), chuy'ywo'm on (1463 Pep 2:207a). Contracted <
chu'i- (chu'i'kwo 1463 Pep 2:241a, chu'ike'tun 1518 Sohak-cho 9:24b), causative of chu- j.
muy- ‘hate’: muy'ti (1459 Wei 9:42a); muy'm ye (1447 Sek 13:56b); muyl (1462 'Nung 8:30a);
muysya.m i (1463 Pep 2:19b, but part of a longer passage in which all of the dots are absent);
76 PART I
A Reference Grammar of Korean
muy'nwola (1481 Twusi 7:20b, 23:23a); muy'ywo'm i (1462 ‘Nung 4:27b), muy'ywu'm ul (1462
‘Nung 9:109a, 1464 Kumkang 79b)
suy- ‘be sour’: suy'Gwo (1481 Twusi 15:21b); suy'm ye (1462 'Nung 5:37b, 1463 Pep 6 : 68 b); suyn
(1445 'Yong text 5:4b, 1462 'Nung 2:115b, 1466 Kup 1:32a); suyl (1527 Cahoy 3:6b = 14a)
ptuy- ‘make/let it float’: 'ptuy'Gwo (1481 Twusi 10:34b); ptuy'ywo'm on (1462 'Nung 6:26b),
ptuy'ywu'm i (1463 Pep 7:50a)
stuy- ‘wear (a belt), gird oneself with’: stuy'Gwo (1482 Nam 2:18b); stuy'm ye (1586 Sohak 2:2b);
stuy'ye (1463 Pep 2:39b [broken type]); stuy'sya (1445 'Yong 112); "stuy < *stuy-'i ‘belt’ (der n)
poy-\ ‘get pregnant with (child)’: poy'Gwo 'eye (1462 'Nung 7:55b); poyn (1463 Pep 6:47a); poyl
(1527 Cahoy l:17b=33b); poy'ya (1447 Sek 13:10a, 1462'Nung 4:76a); poy'ywon (1459 Wei 8:81a)
poy- 2 ‘soak’: poy'Gwo (1462 'Nung 5:88a); poy'n i (1466 Kup 1:16a)
moy- ‘tie/sew (on), attach’: moyl (1459 Wei se:3a, 1462 'Nung 8:106b); moy'n i (1449 Kok 76);
moy'ya (1463 Pep 4:37b), moy'ye sye (1481 Twusi 8:53b) - Cf moy'ye (1465 Wen 1:1:1:89b) <
moy'i- (VP), moy'Gye (1459 Wei se:3b) < moy'Gi- (VP); moy'syan (1462 'Nung 5:24a), moy'yesi'na
(1481 Twusi 7:34a); moy'ywom (1465 Wen 1:2:2:161a), moy'ywo.m i'la (1462 ‘Nung 5:88a),
moy'ywo'm ol (1464 Kumkang 83, 1465 Wen 1:1:1:101b); moy'ywolq (1462 'Nung 7:8a), moy'ywon
( ? 1468- Mong 58, 1481 Twusi 8:47b)
ptwuy- ‘jump’: ptwuy'Gwo (1482 Nam 2:66a); ptwuy'm ye (1482 Kum-sam 2:13a); ptwuyl (1459
Wei 9:19b); ptwuy n i (1482 Kum-sam 4:31b); ptwuy'ye (1481 Twusi 16:2b; 1462 'Nung 8:15a,
8:40a, 8:139a); ptwuy'ywo'la (1462 'Nung 8:139a)
pthwuy- ‘spring, snap, splash’: pthwuy'n i (1481 Twusi 25:53a); pthwuy'nwos.ta (1481 Twusi
25:19a; dot smudged); pthwuy'ye (1481 Twusi 17:13a). VC pthwuy Gwun (1445 ‘Yong 48).
twoy- < towoy- < toWoy- ‘become’: twoy'Gwo (1518 Sohak-cho 8:3b; faint dot), twoy'kwo (1586
Sohak 1:7b); twoy'n i ’la (1518 Sohak-cho 10:6b); twoy'sikwo (1588 Cwungyong 19b). The unattested
infinitive should be *twoy'ya; the modulated stem *twoy'ywo~.
2.12.4.3. Vowel-final monosyllabic stems that are high.
Grouji 3. Stems that are high in all forms
'skwu- ‘dream’: 'skwu'kwo (1449 Kok 67); 'skwum 'skwulq (1462 'Nung 4:130a); 'skwumywo'm i
(1475 Nay 2:2:73a)
'hwo- ‘broad-stitch’: 'hwol (1527 Cahoy 3:9a= 19b); 'hwo'wa [= hwo'a] is'kwo ( ? 1517- 'No 2:52b)
'twoy- ‘measure’ - noun 'twoy ‘(measure)’ (1459 Wei 9:7b); 'twoy'ti ( ? 1517- Pak 1:67b);
'twoy'Gey ( ? 1517- Pak 1:12a); twoy'm ye n’ ( ? 1517- Pak 1:12a); twoy'non (1459 Wei 9:7b);
twoyGe'nul (1447 Sek 6:35b); 'twoyte'n i (1447 Sek 6:35a, 1449 Kok 168); 'twoyywo'm i'lwoswo'n
i (1481 Twusi 8:10a). The unattested infinitive should be *'twoy'ya.
'pwuy- ‘cut’: 'pwuyGwo (1481 Twusi 8:61a) - "pwuyGwo (id. 7:38b) must be a mistake; 'pwuyn.i
(1481 Twusi 7:38b); pwuyl (1527 Cahoy 3:3a=5b) = 'pwuylq (1459 Wei 8:98b); 'pwuyye (1481
Twusi 7:32b) = 'pwuy'ye (1459 Wei 1:45a, 1482 Kum-sam 4:31a); pwuynon (1481 Twusi 7:18b,
10:32a); 'pwuy'ye'tun (1459 Wei 1:45a). The modulated form in pwuyywo'm ol (1481 Twusi 21:24b)
should carry a dot at the beginning; I lack access to the text of 0)'pwuy'ywul (1481 Twusi 9:30_).
'phwuy- ‘burn (a fire)’: phwuy'Gey (1462 'Nung 7:16b, 7:18a); phwu'm ye (1459 Wei 7:35a);
'phwuyn (1462 'Nung 7:18a). Presumably the infinitive was *'phwuy'ye. The causative stem is
phwuy Gwu-.
'chu-i ‘eliminate, get rid of’: 'chu'key (1463 Pep 2:214b); 'chu m ye (1465 Wen 2:1:1:52a); chul
(1465 Wen se:47a); 'chwu'm (~ on 1459 Wei 13:21a, ~ un 1463 Pep 2:207a)
'chu -2 ‘sift’: che (1459 Wei 17:17b, 1462 'Nung 7:9a); 'chwu'm on (1463 Pep 5:155b)
chu- 3 ‘dance’: 'chukwo (1481 Twusi 8:41b); 'chu'mye (1459 Wei 21:190b)
'khu- ‘big’: 'khu'kwo (1447 Sek 6:32b); 'khu'kuy (1447 Sek 6:34a), 'khu'key (1462 'Nung 1:3a,
1518 Sohak-cho 9:24a); 'khu'm ye (1462 'Nung 2:4ab, 1463/4 Yeng 2:12b), khu'm ye n’ (1459 Wei
23:77a); khul (1527 Cahoy 3:1 la = 25b); 'khun (1445 'Yong 27; 1462 'Nung 4:18b; 1463 Pep
A Reference Grammar of Korean
PART I 77
2:190a, 2:231b, 2:232a, 7:141b; ? 1468- Mong 47b; 1482 Kum-sam 3:25b, 4:22a); 'khe (1447 Sek
6:12b, 1449 Kok 28, 1459 Wei 2:47b), 'khe 'za (1463 Pep 2:224b); 'khu'kenul 'za (1482 Kum-sam
2:16a); 'khwu.mu'lwo (1459 Wei 1:29b)
phu- ‘bloom’: 'phu'kwo (1459 Wei 21:2a); phulq (1447 Sek 13:25a); phu'm ye (1459 Wei 2:31a,
21:6b), phu.m ye (1482 Kum-sam 2:6b), 'phu'm ye n’ (1459 Wei 2:47a); phun (1447 Sek 13:25a,
1481 Twusi 21:15b), phu.n i ’’la (1459 Wei 2:47a); 'phe (1449 Kok 158, 1459 Wei 2:47a), 'phe
’ys.non (1482 Nam 1:37b), phe ’ys.ke'tun (1463 Pep 6:47a), phe ’ys.te'n i (1447 Sek 6:31a, 1449
Kok 9, 1459 Wei 1:21a, 1481 Twusi 8:34b), phe ’y'sywo.m ol (1481 Twusi 23:30b); phuke'tun
(1459 Wei 8:75b); 'phutwo'ta (1482 Kum-sam 3:33a, 1482 Nam 1:66a); 'phu-'tos (1463 Pep 1:85b);
phwu'm i (1462 'Nung 1:19a), 'phwu.m ol (1482 Kum-sam 3:33b); 'phwulq t i (1462 'Nung
l:19ab); [Jphwu'toy (1459 Wei 7:57b)
'ptu-i ‘float’: 'ptu'ti (1462 ‘Nung 6:26b), ptun (1462 ‘Nung 1:62b; 1481 Twusi 7:12b, 21:22b);
ptu'lak (1481 Twusi 7:2a); ptu'mye (1462 'Nung 3:79b) and ptu.m ye (1462 'Nung 2:31a); 'pte
(1462 ‘Nung 1:47b); ptwu'mi (1462 ‘Nung 3:106a)
ptu -2 ‘open (eyes)’: 'ptu'kwo (1449 Kok 65); ptukuy (1482 Kum-sam 2:59b); pte (1462 ‘Nung
1:59a); 'ptuke’na (1459 Wei 8 : 8 b) 'ptwu'm i (1463 Pep 2:163b)
ptu -2 ‘spoil’: ptul (1527 Cahoy 3:6a=12a); ptun (1466 Kup 2:61b); ptuno'n i (1466 Kup 2:61b)
'pthu- ‘burst’: pthuti (1463 Pep 2:243a); 'pthukwo (1481 Twusi 25:26b); pthe (1481 Twusi 7:24b)
psku- ‘extinguish’: pskuti (1482 Kum-sam 5:3a); 'pskun (1459 Wei 8:38b, 1463 Pep 6:153a);
pske (1459 Wei 2:71b, 1462 ‘Nung 2:43b, 1481 Twusi 25:13a); 'pskuke'nul (1447 Sek 6:33b);
pskunun (1449 Kok 106); psku'sya (1449 Kok 101); 'pskwutoy (1482 Kum-sam 5:3a)
'psu -1 ‘bitter’: 'psuta (1481 Twusi 8:18a); 'psu'm ye (1462 ‘Nung 3:9b, 5:37b); psun (1462
'Nung 3:9a, 1466 Kup 78b, 1482 Kum-sam 2:50a); pse (1459 Wei 2:25b); pswum (1462 'Nung
3:10a). The expected dot is mistakenly omitted in psul (1527 Cahoy 3:6b= 14b).
'psu -2 ‘use’: psu ti (1447 Sek 19:30b), 'psuti (1482 Kum-sam 5:8a) - 1518 Sohak-cho 10:1b has
psu'ti non but that may be a mistake, since the preceding line has psul\ there seems to be a mistake
also in psu'key (1459 Wei 23:73a) ‘so as to use’, countered by psukwo ’ la a little later in the same
passage; pse (1462 'Nung 1:81a, 1463 Pep 2:240a, 1481 Twusi 8:17a); psunon (1451 Hwun-en lb,
1482 Kum-sam 2:17b); 'psu'sya (1445 ‘Yong 77); 'pswu'toy (1464 Kumkang 87b), 'pswu.m i (1482
Kum-sam 5:8a); pswul (1462 'Nung 1:19a)
(s)su-i ‘write’: sun (1482 Kum-sam 3:7b); ssul (1527 Cahoy 3:9a=20b); 'ssu.m ye (1447 Sek
6:43a, 1463 Pep 2:163a); sse (1447 Sek 9:30a), 'se (1481 Twusi 23:44a); 'sswutoy (1463 Pep
4:72b), 'sswun (1447 Sek se:4b)
'(s)su -2 ‘wear on head’: 'sukwo (1482 Nam 1:30b, 1481 Twusi 15:6b), 'ssu'kwo (1459 Wei
10:95b, 1463 Pep 7:176a)
stu - 1 ‘cauterize’: 'stukwo (1466Kup 1:22a); stu'l i (id. 1:41a); stula (id. 1:36b) = stu'la (id. 1:3a,
19a, 25a, 26b, 29a, 76a); 'stu'm ye n’ (id. 1:22a); 'stum ( ? 1517- Pak 1:38a); 'stu'n i (ibid.); ste (1466
Kup 2:72b); 'stwutoy (1466 Kup 1:20a, 36b) = 'stwu'toy ( ? 1517- Pak 1:38b); stuno'n i (id. 1:57a)
'stu -2 ‘scoop’: stu kwo (1481 Twusi 15:54a), 'ste (1475 Nay 1:3a), 'stul (1527 Cahoy 2:15a=7a)
'cho- x ‘cold’: 'cho'kwo (1459 Wei 1:26b); chol (1527 Cahoy l:lb= la, 3:lb=2a); 'chon (1449
Kok 102); 'chwom (1462 'Nung 3:12a). The unattested infinitive should be *'cha.
'cho - 2 ‘get full’: 'cho'ti (1449 Kok 180); 'cho'm ye (1447 Sek 19:7b); 'chon ( ? 1517- Pak 1:55b,
1518 Sohak-cho 8:27b); cha (1447 Sek 6:4b, 1449 Kok 140), 'cha 'za (1462 'Nung 8:28b); 'choke
za (1447 Sek 19:39a); chwo m ol (1463 Pep 3:98b); cho'sya (1459 Wei 2:8b)
'cho -3 ‘kick’: 'chokwo (1482 Nam 1:50a); thi-'cho'm ye (1449 Kok 39); chol (1527 Cahoy
3:4b = 8 b); pak 'cha (1481 Twusi 15:33a)
cho -4 ‘attach, fasten on’: 'cho'ti (1481 Twusi 8:49b); cha (1465 Wen se: 8 b); 'choke'na (1462
‘Nung 7:46a); 'chwon (1481 Twusi 25:8a)
'pho- ‘dig’: 'pho'kwo (1449 Kok 60); 'pha (1459 Wei 1:7b, 1462 'Nung 7:9a, 1481 Twusi 21:42a);
phwom (1463 Pep 4:95b), 'phwo'm ol (1462 'Nung 3:87b)
78 PART I
A Reference Grammar of Korean
'pto- ‘pick, pluck’: 'ptokwo (1475 Nay 2:2:69b), 'pto.l i ’’Gwo (1481 Twusi 10:8b); 'pta (1449
Kok 99, 1459 Wei 2:12b); ptonon (1475 Nay 2:2:68b, 1481 Twusi 8:15b); 'ptoten (1481 Twusi
15:21a), 'ptwo.m ol, 'ptwotoy (1475 Nay 2:2:69a)
ptho- 1 ‘pluck (harp strings), play (string music)’: 'pthokwo (1481 Twusi 24:38a, 1482 Kum-sam
4:10b); 'ptho.m ye (1482 Kum-sam 2:11b); pthol (1459 Wei 8:49a, 1527 Cahoy 2:9a=17a);
pthono'ta (1482 Kum-sam 5:8a). The unattested infinitive would be *'ptha.
'ptho -2 ‘cut open, split’: 'ptho kwo (1459 Wei 23:73b, 1466 Kup 2:79a); ptha (1482 Nam 1:15a)
'pco- 1 ‘salty’: 'pco'm ye (1462 'Nung 5:37b); 'pcol (1527 Cahoy 3:6b=14a, 3:8a=17b); 'peon
(1459 Wei 1:23a, 1466 Kup 1:32a); 'pewom (1462 'Nung 3:51a). The unattested infinitive: *'pca.
'pco- 2 ‘weave’: 'pco 'ti (1462 'Nung 9:53b), peon ( ? 1517~ Pak 1:29a), pcol (1475 Nay 2:2:51b,
1527 Cahoy 3:8b=19a); 'pea (1475 Nay 2:2:51a, 1481 Twusi 20:19a); 'pewon (1463 Pep 2:140a).
The forms pcoy'isya (1475 Nay 2:2:51b) and pcoy 'ye (id. 2:52a) are from the causative pcoy'i-.
'psko- ‘peel, husk, shell; hatch’: 'psko'kwo (1481 Twusi 7:32-3); 'psko'm ye (1463 Pep 2:116a,
2:117a). The unattested infinitive would be *'pska, the modulated forms *'pskwo-.
'pso- ‘wrap’: 'pso'kwo ( ? 1517~ Pak 1:28a) 'pso.mye n’ (1481 Twusi 16:67b); 'pson (1481 Twusi
8:33-4); 'psa (1481 Twusi 20:39a); pswon (1481 Twusi 21:4b)
(s)so- ‘valuable’: 'sso'ta ( ? 1517- 'No 2:4b); 'sso'm ye (1447 Sek 13:22b); 'sson (1459 Wei
18:78b); ssa (1463 Pep 2:140a). The unattested modulated forms would be 'sswo- .
tho- 1 ‘ride’: tho’ti ( ? 1517- Pak 1:37b); 'tho’kwo (1459 Wei 10:28a), 'thokwo 'eye s (1481 Twusi
15:55b); 'thol (1459 Wei se:18a); 'thon (1459 Wei 1:27b); 'tha (1482 Nam 1:36b); 'tho'sya (1459
Wei 1:27b); 'thwon (1445 ‘Yong 34)
'tho- 2 ‘receive, undergo’: thoti (1481 Twusi 8:33a); tho.m ye (1475 Nay l:se:2b); thon (1447
Sek 19:2b); tha (1481 Twusi 7:2b); 'tho'no.n i ’ la (1462 'Nung 1:89a); thwo.m i (1475 Nay 3:63a)
tho- 3 ‘burn (a fire):’ 'tho'ti (1462 ‘Nung 9:108b); 'thol (1465 Wen 1:1:2:181a)
'hoy- ‘white’: hoyta (1482 Kum-sam 4:22b); 'hoy'Gwo (1459 Wei 1:23a); hoy'Gey (1445 'Yong
50); 'hoyl (1459 Wei 1:22b); 'hoyn (1445 'Yong 50; 1447 Sek 6:43b; 1481 Twusi 7:1a, 16:60a; 1527
Cahoy 2:14b = 29b); 'hoyGe'nol (1481 Twusi 16:1a). Probably scribal errors: ? /“ Jhoy'Gwo (1463 Pep
1:148b); Y. Jhoyywo'm ol ko'cang muy'nwola (1481 Twusi 23:23a) - CF 'hoyywo'm ol (1481 Twusi
7:27a). The unattested infinitive: *'hoy'ya. Variant huy- ( huyn mo "toy 1481 Twusi 25:2b).
moy- ‘remove (weeds), weed’: moyl (1527 Cahoy 3:3a=5a); moyya (1481 Twusi 7:34b);
'moy'ywo'm i (1462 'Nung 1:19a); 'moy'ywolq t i (1462 'Nung 1:19a)
'soy- ‘leak’: 'soy'ti (1463 Pep 3:56a); 'soy'm ye n’ (1466 Kup 1:78a) - f'Jsoy'm ye n’ (1459 Wei
23:77b) omits the dot here and twice above in the line; 'soyl (1463 Pep 1:24b); 'soy'ya (1463 Pep
6:89b); 'soy'non (1462 'Nung 6:106b, 1465 Wen 1:1:2:107b), 'soynon (1447 Sek 13:10b); 'soy'ywom
(1465 Wen 1:1:2:97b)
'skoy- ‘wake up’: 'skoyti (1481 Twusi 10:7a, 1485 Kwan 3a); 'skoy'Gey (1459 Wei 13:18b);
skoy'm ye (1464 Kumkang 38a, ? 1468 Mong 42b); 'skoylq (1447 Sek 9:31a); skoyn (1465 Wen 1:2:
1:47a); 'skoy'ya ( ? 1468“ Mong 59a); 'skoy'yan (1465 Wen 1:1:2:151a); 'skoyGe’na (1459 Wei
10:70b); 'skoyywo'n i (1459 Wei 10:24b), 'skoy'ywom (1465 Wen 1:1:2:37b), skoyywo'm ol (1481
Twusi 21:20b).
'chuy- ‘slant, lean’: 'chuyti (1463/4 Yeng 1:52a), 'chuyn (1459 Wei 1:45b); 'chuy'ye (1465 Wen
1:1:110a, 1586 Sohak 2:62a)
Critical examples are lacking for 'kuy- ‘crawl’, skuy- ‘shun’, spuy- ‘drain’, 'pco- 3 ‘squeeze’, and a
few others. The only examples of 'muy- ‘get cracked’ are of the infinitive, as in 'muyye 'tye (1481
Twusi 16:29b). These are put into the always-high group by default.
'iy- (copula): 'i'Gwo (1459 Wei 1:31a); 'i'Gey (1462 ‘Nung 2:27b); 'i'la (1447 Sek 6:17a;
indicative assertive); - I'm ye (1463 Pep 5:30a), I'm ye n’ (1459 Wei 2:49a); ••• 'in (1462 'Nung
2:6b, 2:8b; ? 1517- 'No 2:54b); Tsya (1447 Sek 13:29a), 'isi'na (1449 Kok 2); i-ngi ''ta (1447 Sek
24:46b, 1459 Wei 21:218b); 'ila'n i (1446 Sek 6:19b; retrospective modifier). Presumably the loss of
accent is secondary in syel'hu.n in 'hoy 'yey (1462 'Nung 2:6b), "twul.h i'm ye (1447 Sek 13:49b),
'SYANG isya- s-ongi ’’ta (1447 Sek 23:22b), ... .
A Reference Grammar of Korean
PART I 79
'chi- ‘raise’: chi'kwo (1463 Pep 7:77b), chil (1459 Wei 8:87a); 'chinon (1459 Wei 1:46b)
'pski- ‘insert, ... ’: pskikwo (1481 Twusi 10:26a) = pski'kwo ( ? 1517- Pak 1:26a); pskil (1459
Wei 13:56b), pski m ye (1463 Pep 5:13a); 'pskye (1459 Wei 2:18b, 1465 Wen 2:3:1:54b, 1475 Nay
se:7a) - "pskye (1466 Kup 1:88a) must be a mistake
pti- ‘steam’: 'ptil (1462 'Nung 4:18b, 1527 Cahoy 3:6a= 12a); ptin (1481 Twusi 7:18a, 1586
Sohak 5:48b); ptye (1462 'Nung 6:89b, 1481 Twusi 20:38a, 1482 Kum-sam 5:45b); ptinon (1481
Twusi 8:9b)
'thi- ‘hit’: 'thi'm ye (1462 'Nung 8:88b), thi'm ye n’ (1447 Sek 6:28a); thil (1527 Cahoy 3:
13a= 30a); 'thye (1447 Sek 6:28a, 1449 Kok 156); 'thike'tun (1459 Wei 7:53b); 'thino'ta (1462 'Nung
4:130a); thywo'toy ( ? 1468- Mong 53a), thywon ( ? 1468 _ Mong 10a), thywum (1463 Pep 5:38a)
2.12.4.4. Vowel-final monosyllabic stems that are high/low.
Group 4. Stems that are high only before the infinitive ending or one of the bound stems
°ci-i ‘want to do’ (aux): ci'la (1447 Sek 24:8a, 24:9b; 1459 Wei 1:10b, 1:11b, 7:12a, 8:101b; 1462
Pep 2:28b; 1481 Twusi 8:1b, 22:35a) II 'd'ye (1462 'Nung 1:16b), 'eye (1447 Sek 6:14b, 6:15a;
1451 Hwun-en 3b; 1459 Wei 18:3a, 21:124-5; 1462 'Nung 1:38a, 7:73b; 1465 Wen se:43b, 1:1:2:75b,
2:3:1:47a; 1481 Twusi 7:14a, 8:38b, 15:55b; 1586 Sohak 6:35b), 'eye ’y.n i ’'la (1463 Pep 4:
134b); ci-ngi ’'ta (1445 ’Yong58; 1447 Sek 6:22b, 24:8b; 1459 Wei 2:9b, 2:27b, 8:1a, 8:4-5,10:10b)
°ci- 2 ‘carry on the back’: ci'kwo (1463 Pep 2:165a), cil (1527 Cahoy 3:10b = 24a); "cywum (1463/4
Yeng 2:73b), "cywu'm ul (1481 Twusi 24:32a) II 'eye ( ? 1517- Pak 1:11b; 1481 Twusi 7:28a) < 'd'ye
°d- 3 ‘chop (wood)’: "cywu'm ey (1481 Twusi 7:6b) II 'cinon (1481 Twusi 7:39a). This
assignment assumes such unattested forms as *ci'ti, *ci'kwo, *cin, *cil(q), *'cye < *'ci'ye.
°(c)ci- ‘get fat’: ci'ti ( ? 1517- Pak 1:22b); ci'bA’o (1481 Twusi 16:62b); cin (1481 Twusi 15:4b,
1466 Kup 1:80a); cci'key (1459 Wei 23:73a) II solfhj 'eye (1481 Twusi 16:15b) < *'ci'ye
°i- ‘carry on the head’: i'ta (1482 Nam 2:64a); i'kwo (1482 Kum-sam 2:11a - also 1481 Twusi
18:10, unavailable to me); il (1527 Cahoy 2:10b = 24b) II iye (1449 Kok 34) = 'i'ye (1462 'Nung
8:93b, 1463 Pep 4:174a). The substantive *i'm ye is unattested; "im (1482 Kum-sam 5:34a) is the
modulated substantive (we expect * "ywum < *i'ywum ) = 'iywo'm on (1482 Nam 2:64a), see p. 70.
°ni- ‘roof, thatch’: "nil = nil (1527 Cahoy 3:8r=18r) II 'nisi'kwo (1475 Nay 2:2:72b). See p. 70.
°ti-i ‘fall’: ti'kwo (1445 'Yong 86); ti'key (1459 Wei 1:29a); til (1527 Cahoy 3:3a=5a); tin t ol
(1445 'Yong 31) - “'tin tof (LKM 1962:117) must be a misprint (Cf Taycey-kak repro) II ti ye
(1447 Sek 9:27b, 1518 Sohak-cho 10:11b), 'tiye (1481 Twusi 15:44a); tike'nul (1447 Sek 6:30-1),
'tike'tun (1462 'Nung 1:19a); 'tinwon (1481 Twusi 21:14b); "tywu.m ul (1481 Twusi 23:30b),
"tywu'm i (1482 Kum-sam 2:49b), "tywu.m ay (1482 Kum-sam 2:6b); "tywulq (1447 Sek 9:28a)
°ti- 2 ‘become’ (aux): ti'key (1447 Sek 6:13a); ti'kwo (1481 Twusi 20:16a); dm ye (1459 Wei
2:71b); II 'd'ye ’ys.ke'nol (1481 Twusi 15:44b)
°psd- ‘overflow’: psti'm ye (1449 Kok 178), psd'kwom (1459 Wei 7:9b). There are no attested
examples that would call for psti- , but such forms as psti'ye and psti(')ke- must have existed.
°ca- ‘sleep’: ca'd ( ? 1517- 'No 1:47b); ca'kwo ( ? 1517- 'No 1:10b); ca'key ( ? 1517- 'No 1:46b); cam
ye ( ? 1468- Mong 42b); calq (1462 'Nung 9:88a), cal (1459 Wei 1:25b, 1527 Cahoy l:15b = 30b) II
ca za (1481 Twusi 16:66a); 'casya (1482 Nam 2:76a), 'cano'n i (1447 Sek 13:10b)
°ha- ‘many/much’: halk ka [= halq ka] (1465 Wen 1:2:2:136a); ham ye (1459 Wei 10:19a),
ha'm ye n' (1481 Twusi 22:20a); han (1445 'Yong 19, 1447 Sek 6:25b, 1459 Wei 17:44a, 1463 Pep
4:84b) - 'han 'pi (1445 'Yong 67) ‘heavy rain’ must be either a mistake or a variant of the
modulated modifier "han (1447 Sek 6:2b); han ye (1447 Sek 19:4a), han i ’’la (1459 Wei 2:31b)
II 'ha 'two (1463 Pep 7:62b), ha'a (1459 Wei 1:24b), 'hano'n i (1445 'Yong 2, Manlyek text);
'hasin (1449 Kok 18); "ha'm ol (1482 Kum-sam 3:19a) *- ha-'[wo]-m. See p. 70 for ha-ngi ’’ta.
°ka- ‘go’: ka'ti ( ? 1517- Pak 1:67b, ? 1517- ‘No 2:7a); ka'kwo (1462 'Nung 7:73b); ka 'key (1447 Sek
6:9b); ka'm ye (1459 Wei 8:10b, 1481 Twusi 7:3b), ka'mye n’ (1447 Sek 6:22b); ka-'tos (1459 Wei
2:7a) II’ka (1447 Sek 6:35b; 1459 Wei 2:11a, 10:20b, 18:71b; 1462 'Nung 2:50b; 1463 Pep 1:77a;
80 PART I
A Reference Grammar of Korean
1481 Twusi 7:2a, 8:37b, 8:40a; ? 1517- Pak 1:37b, 1:54a, 1:64b); 'ka'a (1447 Sek se:6b, 6:1a, 6:6b,
13:10b, 24:37b; 1459 Wei 8:100b, 10:13ab, 18:71b, 23:65a; 1463 Pep 2:138b); 'ka'sya (1445 'Yong
58; 1447 Sek 6:45b; 1459 Wei 1:5b, 2:11b), ' ka'non 'ta ( ? 1517- 'No 1:1a), kano'n i (1445 'Yong 2,
1447 Sek 6:9b), 'kanon ce'k uy (1447 Sek 6:19a), ‘kata ‘ka (1445 'Yong 25, 1482 Nam 1:36b)
°na- ‘emerge’: na'ti (1462 'Nung 1:8b) - Cf modulated “na'ti (1447 Sek 6:19a); na'kwo (1459
Wei 1:46a); na ‘key (1463 Pep 1:158b); nan (1459 Wei 1:28b, 21:216a; 1463 Pep se:7b); nal (1527
Cahoy l:17b = 34a); na'm ye (1447 Sek 23:44a; 1462 'Nung 1:51b), na'm ye n’ (1462 'Nung 7:74b);
"na'toy (1447 Sek 19:7b, 1449 Kok 185) *- na-‘[wo]-'toy, "nalq (1462 'Nung 3:24b) *- na-'fwo]-lq
(1459 Wei 21:215b) II ‘na (1481 Twusi 7:39a); 'na'a (1449 Kok 41, 1459 Wei 1:5a); 'na'taka ’mye
(1459 Wei 21:215b)
°sa- ‘buy’: sa'kwo (1481 Twusi 7:21a); sa'key ( ? 1517- Pak 1:2a), sal (1527 Cahoy 3:9b=21a)
II ‘sa ( ? 1517- 'No 2:21a); ‘sa'a (1447 Sek 6:8a, 1459 Wei 1:10b)
°(h)hye- ‘pull, drag’: hhye'kwo (1449 Kok 39), hye'kwo (1463 Pep 4:93b); hye'ti ( ? 1517- 'No
2:31a); hhyen (1463 Pep 2:100b) - "hhyen (1462 'Nung 1:17b) is the modulated modifier; hhyel
(1459 Wei se:3a), hyel (1527 Cahoy l:18b=35b) II ' hhye (1463 Pep 1:158b, ? 1468- Mong 58a);
hye'a (1482 Kum-sam 2:64b); ‘hyeta 'ka (1481 Twusi 16:1b)
°nye- ‘go’: nye'ti ( ? 1468- Mong 41b, 1481 Twusi 7:6a); nye'key (1463 Pep 2:39b, 1481 Twusi
7:6a); nyel (1447 Sek 9:21b, 1482 Nam 1:28b) = nyelq (1459 Wei 21:119a), nye'l i ‘Gwo ( ? 1517-
*No 1:30b) > nyey.l i Gwo (1795 'No-cwung [P] 1:27b), nye'cye ( ? 1517- 'No 1:10b) = nyeycya
(1795 'No-cwung [P] 1:9b); nye'm ye "nye'm ay (1482 Kum-sam 4:2a) II nye (1481 Twusi 7:2a,
14:29b, 25:29a, 14:29b; ? 1517- 'No 1:1b); nye'a (1449 Kok 86); ‘nyeke'tun (1463 Pep 3:155b);
‘nyenun (1459 Wei 7:52b), ‘nyenon (1481 Twusi 21:14a), ‘nyenwon t ol (1482 Kum-sam 4:2a);
'nyesi'n i (1482 Kum-sam 4:54a), nyesil (1463 Pep 2:39b)
°phye- ‘spread it’: phye'ti (1481 Twusi 8:4b); phye'kwo (1462 'Nung 9:88a); phye'key (1459 Wei
18:61b); phyel (1459 Wei 21:4a, 1527 Cahoy 3:6a=12b); phye'm ye (1481 Twusi 16:55a) II 'phye
(1447 Sek 9:21b, 9:29a; 1462 ‘Nung 1:4a), 'phye'a (1447 Sek 6:6a, 13:10a), phye'e ’ys.ten (1447
Sek 6:2a); 'phye'ta'la (1463 Pep 4:170a); phyesi'm ye phyesi'kwo k (1462 'Nung 1:108b);
'phye'sya (1462 'Nung 1:3a); 'phyesyan tila (1482 Kum-sam 5:35b); phyesi'nwon (1482 Nam 1:5a)
°sye- ‘stand’: sye'ta (1447 Sek 19:13a); i'le sye'ti (1475 Nay 1:34a); sye'kwo (1463/4 Yengka
2:12a), syel (1527 Cahoy 3:12a=27a), "syelq (1462 'Nung 3:36a; modulated) II 'sye'a (1447 Sek
19:31a, 1459 Wei 2:64b), 'sye'e (1459 Wei 10:17b);'.sye'jy a (1445 'Yong 28). The phrase ans.ke'na
['Jsyeke'na (1447 Sek 19:5b) ‘whether sitting or standing’ suppresses the stem accent even before the
effective aspect - ‘ke-, but that is peculiar to this idiom.
°wo- ‘come’: wo'ti (1459 Wei 7:29b); wo'key (1447 Sek 6:43b), wo'kwo (1481 Twusi 16:65a);
wolq t ol (1449 Kok 147), won ‘ta ( ? 1517- Pak 1:51a), wo’n i (1459 Wei 1:45a); wo'm ye n’ (1586
Sohak 4:33a); "wo'm i (1459 Wei 9:10b, 1482 Kum-sam 3:19a), "wo rn ol (1482 Nam 1:50b),
"wo[']m ay (1481 Twusi 21:25b) *- wo-‘[wo]-m II ‘woke'na (1459 Wei 9:43a); ‘woke'nol (1481
Twusi 8:40a); woke'ton (1482 Kum-sam 3:27b), 'woke'tun (1459 Wei 10:25a), 'wona'ton (1447 Sek
6:16b, 19:6a), 'wa'ton (1463 Pep 3:2b); 'wo'sya (1459 Wei 8:55b), wosya 'two (1447 Sek 6:4b),
'wosya 'za (1445 'Yong 38); 'wosi'n i (1459 Wei 1:5b); 'wo'silq (1459 Wei 2:18b), wosil ss ye
(1447 Sek 23:29a); 'wosi'n i (1459 Wei 1:5b); 'wosin 't i (1463 Pep 5:119b); 'wona two (1481 Twusi
25:23a); 'wona 'two (1481 Twusi 25:23a); 'wona'ta ( ? 1517- Pak 1:3a); 'wo'nan t i ( ? 1517- 'No
1:68b); 'wo'na'la (1459 Wei 7:7b, ? 1517- 'No 1:57b); 'wona'n i (1463/4 Yeng 1:90b); 'wonan t i
(1463 Pep se:21 a); 'wona'n ywo (1447 Sek 6:19b); 'wo'no-ngi ’'ta (1447 Sek 6:29b); 'wono'n ywo
(1447 Sek 6:29b); wo'nwo'la ( ? 1517- 'No 1:1b); 'wonwon t in t ay n’ (1459 Wei 10:7b); 'wota 'ka
(1447 Sek 23:57b); 'woten 't ey n' (1445 'Yong 51)
°pwo- ‘see’: pwo'ti (1462 'Nung 2:37a; 1481 Twusi 7:29a, 8:24a); pwo'kwo (1447 Sek 6:14a, 6:
19a, 6:30a, 19:10a, 24:20b; 1459 Wei 17:17b; 1482 Kum-sam 2:1b); pwo'key (1447 Sek 13:10a);
pwon 'ta (1462 'Nung 2:8b, ? 1468- Mong 58a); pwolq (1462 'Nung 2:111a); pwo'm ye (1447 Sek 23:
22a, 13:23b, 19:10a; 1465 Wen 1:2:1:39b); "pwom (1462 'Nung 2:84a) *- pwo-'[wo]-m\ pwo-'two
(1447 Sek 24:28b) II 'pwo'a (1459 Wei 10:4b), 'pwoa 'eye (1447 Sek 6:14b); 'pwoken t ey n’ (1447
A Reference Grammar of Korean
PART I 81
Sek 6:6a), 'pwoken 't ay n’ (1459 Wei 7:12b, 1462 'Nung 2:6-7); 'pwo'asi'tun (1459 Wei 2:58b);
'pwoa'ton (1447 Sek 6:15b); 'pwosi'kwo (1447 Sek 6:17b, 1449 Kok 43); 'pwosi'm ye (1475 Nay 1:
9-10); 'pwosin t ay (1449 Kok 49); 'pwo'sya (1463 Pep 5:100a); 'pwono'n i (1447 Sek 13:25ab,
1462 'Nung 1:108b), 'pwo'no.n i ’’la (1459 Wei 21:206a), 'pwonoti 'ta (1462 'Nung 1:83b);
'pwonwon (1459 Wei 2:53a); 'pwonwo'la (1481 Twusi 15:52b, 'pxvonwo[']la 7:11a); 'pwonwon ka
(1449 Kok 2), 'pwonwon t i'm ye (1459 Wei 17:35a), 'pwonwon t on (1475 Nay 1:77b)
°(s)swo- ‘shoot; sting’ (not spelled °pswo- before the early 1500s): sswo'ta (1446 Hwun 24a);
sswo'ti (1481 Twusi 10:26a); swol (1459 Wei 14:61b); the exceptional 'sswo'm ye (1462 'Nung
8 :88b) seems to be a mistaken continuation of the high pitch of the preceding string ca'pu'm ye 'thi'm
ye - , and the dot on 'pswol ‘to shoot’ (1527 Cahoy 3:5a=9a) is also a mistake that a few entries later
is countered by the expected pswol ‘to sting’ (id. 3:5a=10a). II 'swoa (1481 Twusi 7:18a, 1482
Kum-sam 4:52a), 'swoa 'za (1465 Wen 1:1:1:113a), 'swa (1481 Twusi 16:56b); 'sswo'sya (1445
'Yong 63), 'sswosi'n i (1445 'Yong 57). Also swo'ta (1446 Hwun 24a) ‘overturn’ ?< ‘shoot it down’.
°cwu- ‘give’: cwu'kwo (1463 Pep 4:37b); cwu'key (1465 Wen 2:3:1:125a); cwul (1527 Cahoy
3:9b = 21b, 3:10a=21b), cwu'l i ’Ge n i (1447 Sek 9:13a), cwu'l i 'ye (1447 Sek 9:12b); cwu'mye
(1447 Sek 9:12a), cwu'm ye n’ Pak 1:43a) II 'cwue (1481 Twusi 7:23b), 'cwue 'two (1463
Pep 2:77a); 'cwusi'm ye n’ (1447 Sek 23:55b); 'cwue'nul (1459 Wei 17:20a); 'cwu'esi'tun (1447 Sek
6:22b), 'cwu'esi'ton (1475 Nay 1:9-10); 'cwunu'n i (1464 Kumkang 21b); 'cwu'sya (1445 'Yong
41). Unexplained: "mwut.cye 'cwukwo 'za ( ? 1517- 'No 1:51b) *- *"mwut'cye cwu'kwo za.
°twu- ‘put away’: twu'ti (1482 Kum-sam 2:65a); twu'kwo (1447 Sek 6:23ab, 6:26a, 9:14a; 1459
Wei 1:28a, 21:78b), twu'kwo n’ ( ? 1517- 'No 1:43b, 1518 Sohak-cho 8:37b), twu'kwo 'za (1459 Wei
7:9a); twul tta [= twulq 'ta] (1459 Wei 2:64a), twu'l i Ga (1481 Twusi 8:3b) II 'twu'e (1447 Sek
6:26a); twuten t ay n’ (1463 Pep 2:231b); twusi'kwo (1445 'Yong 58); twu-’sywo'sye khe'nul
(1445 'Yong 107)
°nwu- ‘void (urine/feces)’: nwu'mye n’ (1466 Kup 1:11b) II nwu'non (1586 Sohak 4:30b)
°ho- ‘do’: ho ti (1475 Nay 1:70b); ho'kwo (1447 Sek 6:6a, 6:29a, 6:35b, 13:36a, 24:3b; 1451
Hwun-en 3b; 1459 Wei 1:13b, 1:26b, 1:30a, l:30ab, 2:11a, 2:69a, 7:5b, 7:15b, 7:16a, 8:38b, 9:10b,
9:55ab, 10:9b; 1462 'Nung 8:104b; 1475 Nay 1:9-10, 1:34a, 1:84a; 1481 Twusi 8:27b, 23:23a; 1482
Kum-sam 2:6b, 2:21a, 2:55a, 2:65a, 3:55a, 4:48b; ? 1517- Pak 1:25a, 1:39b, 1:43a, 1:57a; ? 1517- 'No
2:54b; 1586 Sohak 6:9b), ho'kwo k (1481 Twusi 8:33-4, 15:5a), ho'kwo n’ (1459 Wei 17:54a; 1463
Pep 6:15b; 1482 Kum-sam 2:37a), ho'kwo ’'la (1447 Sek 6:46a; 1459 Wei 1:13b, 10:4b); ho'key
(1447 Sek 24:3a, 1459 Wei 21:219b), ho'kuy (1447 Sek se:6a, 9:5a, 24:2b);
ho'm ye (1447 Sek 9:12a, 9:17b, 13:22b, 13:23a, 19:7a, 19:7b, 21:68b, 23:34b, 24:28b; 1459 Wei
2:16a, 2:53a, 10:20b, 21:120a, 21:146a; 1462 'Nung 1:113a, 2:8b, 2:20b; 1463 Pep 3:178b, 5:212b;
1463/4 Yeng 2:126a; ? 1468- Mong 62ab; 1475 Nay 1:76-7; 1482 Kum-sam 2:5b, 2:7b, 3:3b, 5:40b;
? 1468- Mong 12a, 62ab, 1481 Twusi 7:31b; 1586 Sohak 2:9b), ho'm ye n’ (1447 Sek 24:6b, 1459
Wei 1:12b, 1:49b, 8:62b, 10:18a, 18:18b; 1462 'Nung 1:77b, 3:47b, 5:85b, 7:73b; 1463 Pep 4:75a,
1463/4 Yeng 2:70a, 1464 Kumkang 64b, 1466 Kup 2:64a; 1482 Kum-sam 2:5b, 5:48-9);
hoi (1481 Twusi 7:20b, 8:4b, 15:47b; 1459 Wei 1:18a, 18:13b; 1464 Kumkang 81b, 87b; 1475 Nay
1:35b; 1482 Kum-sam 2:20a; 1518 Sohak-cho 8:13b; 1586 Sohak 2:9b), holq (1459 Wei 8:69b); hoi
'tta (1463 Pep 4:176b) = hoi 'ta ( ? 1517- Pak 1:10a, 1586 Sohak 6:50b); ho'l ye (1463 Pep 3:86a;
1464 Kumkang 69b; 1481 Twusi 21:38a); ho'l ywo (1447 Sek 6:24a, 1462 'Nung 2:81a, ? 1517‘ Pak
1:3a); holq t ol (1462 'Nung 3:68b) = hoi 'tt ol (1463/4 Yeng 1:5b); hoi s (1475 Nay se:6a, 1:34a,
1:77a, 3:61a; 1482 Kum-sam 5:10b); hoi 'ss ol (1462 'Nung 1:29a) = holq 's ol (1462 'Nung 2:61a);
hoi 's i (1481 Twusi 8:1b) = hoi 'ss i (1463 Pep 2:60a), hoi 'ss i'Gwo (1447 Sek 9:37a) = hoi 's
iGwo (1482 Kum-sam 2:20b); hoi 'ss i'la (1447 Sek 6:46a, 1459 Wei 2:66b, 1462 'Nung 1:2b) = hoi
's i'la (1465 Wen se:8b, se:77a); hoi 'ss i'm ye (1459 Wei 2:60a); hoi 'ss i'n i (1459 Wei 2:16a;
1462 'Nung 1:2b, 3:12b) = hoi 's i[']n i (1482 Nam 2:6b); hoi 'ss oy (1447 Sek 6:2a, 13:36a,
19:25b, 24:40a; 1459 Wei 2:60a; 1462 'Nung 9:22a, 10:18a, 21:142b; 1463 Pep 1:164a) = hoi f'Jss
oy (1463 Pep 1:158b) = hoi s oy (1481 Twusi 7:5a); ho'l i (1447 Sek 13:15a; 1459 Wei 9:52a,
82 PART I
A Reference Grammar of Korean
14:31b; 1462 'Nung 1:8b, 1:75a, 7:18a; 1463 Pep 1:208a, 2:6b, 2:28b; 1464 Kumkang 43a; 1465
Wen 1:2:2:4a; ?1468- Mong 10b; 1481 Twusi 7:7b, 22:7b); ho'l i 'za (1463/4 Yeng 2:111a); ho'l i
Gwo (1459 Wei 21:49b, 1481 Twusi 7:7b, 8:29a; 71517- Pak 1:7b, 1:64a, 1:74a) = ho'l i f'JGwo
(1459 Wei 21:49b; 1481 Twusi 22:7b; ? 1517- Pak 1:74a); ho'l i ’’la (1447 Sek 6:1b; 1459 Wei
1:17a, 2:36b, 7:15b, 8:7a, 9:35de, 10:14b; 1462 'Nung 1:44a; 1463 Pep 3:47a; 1475 Nay se:6a);
ho'l i 7 'ss oy (1447 Sek 6:2b; 1459 Wei 1:28a, 2:61a, 7:15b) = ho'l i 7 s oy (1481 Twusi 8:2b);
ho'l i ’ n i (1459 Wei 1:49b, 1462 'Nung 5:85b, 1465 Wen 1:2:3:6a) = ho[']l i ’ n i (1463 Pep
4:86b); ho'l i ’n t ay n’ (1482 Nam 2:6ab); ho'l i ’-ngi ’’ta (1447 Sek 6:4a, 24:28a, 1459 Wei 1:17a,
10:12b); ho 'l i ’-ngi s 'kwo (1459 Wei 9:24a, 23:91b; 1482 Kum-sam 2:4ab); ho'l i 'n t ye (1475
Nay 1:7a); ho'l ye (1462 'Nung 3:43b; 1463 Pep 3:86a); ho'la (1447 Sek 6:9b, 9:41a; 1459 Wei
7:42a, 8:8b; 1481 Twusi 8:7a, 25:56b; 1482 Kum-sam 5:14a; 1482 Nam 1:50b; 71517- Pak 1:6a);
ho'la ’n t oy (1447 Sek 9:26b; 1459 Wei 13:35b, 1462 'Nung 3:24b; 1482 Kum-sam 2:2a);
Unexplained: coy'cwo hoi syeng i (1518 Sohak-cho 8:37b) ‘those who have talent’.
hon (1445 'Yong 47, 123; 1447 Sek 6:15b, 6:19a, 6:22a, 9:19-20, 13:33b, 13:39a, 13:47a, 13:39a,
24:2a; 1459 Wei l:14ab, 1:23b, 1:46a, 2:12a, 7:7b, 7:48a, 8:38b, 10:8b, 10:9a, 10:19a, 17:12v,
18:13b, 21:34ab, 21:129a, 21:216a; 1462 'Nung 1:3a, 1:18b, 1:23b, 1:77b, 1:113a, 2:81a, 2:92b,
2:98a, 2:111a; 1463 Pep se:23a, 2:24a, 2:26a, 2:172ab, 3:180ab; 1464 Kumkang 72b; 1465 Wen
2:3:2:68a; 71468- Mong 20b, 47b; 1475 Nay 1:25b, 2:2:47b; 1481 Twusi 7:23b, 8:13b, 8:42a,
13:13a, 16:61b, 20:29a, 21:3b, 21:20a; 1482 Kum-sam 2:68b, 3:25b, 5:30-1; 71517- Pak 1:64a;
? 1517- ‘No 2:53b; 1586 Sohak 2:9b, 4:13a); ho'n ywo (1459 Wei 8:95a); hon t ol (1445 'Yong 69;
1459 Wei 17:17b, 17:33b), hon ton (1447 Sek 24:18a; 1459 Wei 2:70b; 1481 Twusi 8:7a); hon 't
ay (1447 Sek 24:49b, 1459 Wei 8:101b, 1481 Twusi 24:13a); hon 't oy (1462 'Nung 7:54a); ho'n i
(1445 'Yong 6, 18; 1447 Sek 6:5ab, 6:6a, 6:17b, 6:22a, 9:19-20; 1459 Wei se:lla, 18:7b, 21:216ab,
23:65b; 1462 'Nung. 1:53a, 2:17b, 2:40b, 7:27a; 1463 Pep 1:249a, 3:196b, 6:144a; 1465 Wen
1:1:1:45b, 3:3:1:62a; ? 1468- Mong 20b; 1475 Nay se:8a, 1:18a; 1481 Twusi 7:13b, 8:9a, 15:42b,
15:47b, 16:70b, 22:50a; 1482 Kum-sam 2:1b, 2:2a, 3:3b, 3:19b, 3:34b; 1518 Sohak-cho 10:3a); ho'n
i 'za (1447 Sek 24:20b; 1482 Kum-sam 2:3a); ho'n i ’'la (1447 Sek 6:2b; 1459 Wei 7:44b, 7:70a;
1462 'Nung 1:113a, 4:11a, 10:42b; 1463 Pep 2:113b, 2:173a; 1464 Kumkang 11a; 1475 Nay
2:1:30b; 1482 Kum-sam 2:3a; 1482 Nam 2:5a) = ho'n i ’['Jla (1465 Wen 2:3:1:38b, 1481 Twusi
16:47b; 1482 Nam 2:5a); ho'n i -ngi s kwo (1445 'Yong 28); ho'n i Ga (1481 Twusi 7:14a,
10:42a); ho'n i Gwo (1518 Sohak-cho 10:24b) = ho'n i [JGwo (1481 Twusi 7:40a); ho'n i ya
( ? 1468- Mong 31a, 31b); ho'na (1462 'Nung 2:89b); ho-'tos (1447 Sek 13:45a, 1462 'Nung 1:53a)
II 'ho'ya (1445 'Yong 123; 1447 Sek se:2b, 6:1a, 6:3b, 6:4a, 6:8a, 6:8b, 6:9a, 6:11a, 6:13b, 6:15b,
6:16b; 6:23a, 6:27b, 6:34a, 6:35b, 9:4b, 9:14a, 9:24b, 9:40a, 13:19a, 13:36a, 13:43b, 13:49b,
13:57a, 13:57b, 13:58a, 13:59a, 13:61a, 13:58a, 13:59a, 13:61a, 18:26b, 19:6a, 19:8a, 19:29b,
23:11b, 23:29a, 24:6b, 24:29a, 24:37b; 1459 Wei 1:12b, 1:15a, 1:16b, 1:17b, l:30ab, 1:53a, 2:42b,
2:60a, 2:69a, 7:13b, 7:17b, 7:31-2, 8:7a, 8:104b, 9:52a, 10:31a, 10:31b, 13:35b, 17:35a, 13:43b,
17:54a, 17:85a, 18:3a, 18:7b, 18:26b, 21:20a, 21:120b, 21:129b; 1462 'Nung 1:58a, 1:90b, 2:6b,
2:67a, 3:42b, 6:29a, 7:13a, 7:24a, 7:73b, 10:1b, 10:18a; 1463 Pep se:21a, 1:208a, 2:172ab, 2:226a,
3:47a, 3:104b, 3:196b, 3:197a, 4:75a, 4:154b; 1463/4 Yeng 2:62a; 1464 Kumkang 79b, 87b; 1465
Wen se:5a, 1:2:2:136a; ?1468- Mong 22b, 32b, 43a, 62b; 1475 Nay 2:1:16a; 1481 Twusi 8:27b;
1482 Kum-sam 2:1b, 2:2a; 1485 Kwan 3a, 4b; ?1517- Pak 1:3a, 1:6a, 1:18b, 1:54a; ?1517- 'No
2:19b, 2:36a; 1550 'Yenghem 8b; 1586 Sohak 2:9b); 'ho.ya (?1468- Mong 13b; 1475 Nay 1:77b,
1:84a, 2:2:17b; 1481 Twusi 6:43a, 7:2b, 7:9b, 7:12a, 8:2b, 8:33b, 8:52a, 16:19a, 16:37b, 20:29a,
22:7b, 22:34-5, 23:44a, 25:18a; 1482 Kum-sam 2:10a, 2:24b, 2:57a, 2:65a, 3:27b, 3:31a, 3:50b,
5:35b, 5:40b; 1482 Nam 1:24a, 2:2b, 2:63a), ho.ya 'sye (1481 Twusi 25:56b), 'ho.ya n’ (1481
Twusi 7:29a), 'ho.ya 'two (1459 Wei 1:13a, 21:20a; 1481 Twusi 22:7b, 24:59b), 'ho.ya 'za (1447
Sek 6:2b; 1459 Wei 1:47a, 10:14b; 1462 'Nung 1:44a) = ho'ya za (1586 Sohak 4:9b); ho'ye (1518
Sohak-cho 8:33b), ho'ye 'sye ( ? 1517- Pak 1:54b; 1518 Sohak-cho 8:38-9);
ho.ya nol (1447 Sek 6:8b, 23:23b; 1459 Wei 7:15b; ? 1468 Mong 32ab; 1481 Twusi 22:35a; 1482
A Reference Grammar of Korean
PART I 83
Nam 1:30b); ho.ya'n ywo (1482 Kum-sam 3:52a); ho.yasi'nol (1459 Wei 2:64a, 1482 Nam 1:14a);
'ho.ya'ton (1482 Nam 1:44-5)
hoke tun (1447 Sek 24:3a, 1459 Wei 21:34ab); hoke'na (1447 Sek 13:52a, 1462 'Nung 8:77a);
hoke'nul (1447 Sek 6:16a, 1475 Nay 2:1:30a); 'hokan t i'la (1459 Wei 17:36b), hoka'nywo (1481
Twusi 16:1b); ho'kesi'nol (1459 Wei 2:5a); ho'kesi'n i 'Gwa (1459 Wei 1:12b); hokes'ta (1481
Twusi 21:42a); ho'ke.n i 'Gwa (1482 Kum-sam 3:55a)
'hote'la (1447 Sek 6:15b, 6:30a, 24:3ab; 1459 Wei 2:42b; 1586 Sohak 5:48b), 'hota'la (1447 Sek
6:24b, 1459 Wei 7:14b, ?1517- Pak 1:37b); 'hoten (1447 Sek 6:19a), 'hote'n i (1447 Sek 6:8-9,
6:19a, 19:40b; 1459 Wei 2:42b, 7:24b, 7:29b; 1481 Twusi 7:29a), 'ho'te.n i ’'la (1459 Wei l:8ab);
ho tan (1459 Wei 7:13b); 'ho'tan (1459 Wei 1:7-8, 7:13b, 23:65b), 'hota'n i (1447 Sek 13:57b,
24:3a; 1463 Pep 2:5b, ? 1517- Pak 1:58b), ho'ta.n i ’’la (1463 Pep 1:158b); ho'te-ngi ’’ta (1447 Sek
6:15a), ho'ta-ngi ’'ta (1463 Pep 2:4b); hote'tun (1447 Sek 19:34ab, 1481 Twusi 15:31b);
ho'tesi'ta (1447 Sek 6:la, 6:44a; 1459 Wei 1:18b, 2:26-7); 'hote'sin (1447 Sek 13:58a); ho'tesi'n i
(1449 Kok 41, 1459 Wei 10:18b); 'hotwo'ta (1481 Twusi 7:12ab, 21:15a; 1482 Kum-sam 2:27b,
2:28b; 1482 Nam 1:36a; ? 1517- Pak 1:46b), ho two'ta (1518 Sohak-cho 10:18b), hotwu'ta (1462
'Nung 3:116b), ho'twu'ta (1518 Sohak-cho 10:18b)
hono'ta (1447 Sek 6:2a, 6:14b; 1481 Twusi 7:2a, 8:1b, 8:52a); 'honon (1447 Sek se:la, 9:33a; 1451
Hwun-en lb), ho non (1459 Wei 21:215b, 1462 'Nung 1:77b, 8:104b; 1518 Sohak-cho 9:90a);
honon kwo (1447 Sek 6:27a); honon 'ta (1459 Wei 9:46a, 1462 'Nung 1:84a, 1481 Twusi 8:24a,
?1517- Pak 1:31b); 'ho'non t oy ( ? 1517- 'No 1:35b); 'hono'n i (1459 Wei l:30ab, 5:59b, 9:23-4; 1462
'Nung 6:43a; 1481 Twusi 7:24b, 10:42a; 1482 Nam 1:36b); ho'no.n i ' la (1447 Sek 6:5b, 13:2a;
1459 Wei 1:23b, 2:2a, 10:18b; 1459 Wei 1:11a; 1462 'Nung 6:43a; 1481 Twusi 20:34b; 1586 Sohak
2:25a), = hono.n i ' la (1481 Twusi 16:19b) = hono n i ’la (1482 Nam 2:6ab), 'ho'no.n i non
(1586 Sohak 4:43a), ? 1517- Pak 1:58a); ho'no.n i s ka (1447 Sek 6:18a) = hono.n i s 'ka (1447
Sek 6:16b), hono n i n’ (1481 Twusi 16:39b); 'hono'n ywo (1462 'Nung 2:111a, 1586 Sohak 6:58a);
ho'nwon (1459 Wei 13:35b; 1462 'Nung 1:90b, 2:17a, 7:74b), 'honwon (1447 Sek se:6a, 6:36a,
9:40b, 13:18-9; 1459 Wei se:16a; 1463 Pep 5:212b; 1482 Nam 2:30b; 1586 Sohak 6:44a), 'honwo'n i
(1447 Sek 19:29b; 1459 Wei 1:11b, 13:37a, 21:125b; 1481 Twusi 21:13b), ho'nwo.n i s 'ka (1447
Sek 6:16b); 'honwo'la (1447 Sek 6:8a, 1459 Wei 8:35a, 10:4b, 10:18a; 1462 'Nung 1:17b; 1475 Nay
se:6a, 1:37a; 1481 Twusi 7:5a, 8:35a, 15:23b, 22:26a, 25:23a); ho'nwo-ngi ’’ta (1447 Sek 6:25b);
honwos'ta (1481 Twusi 20:4b, 20:29a)
hosi'kwo (1447 Sek 13:15a, 1459 Wei 10:6a, 21:219a); ho'sike'tun (1459 Wei 8:48b); hosil ss oy
(1459 Wei 2:62b); hosi'ta (1447 Sek 13:30b); ho'sitan (1459 Wei 23:65a); ho'siten ka ( ? 1517-
Pak 1:51a); 'ho'sitas'ta (1459 Wei 21:208a); hosite'la (1447 Sek 13:59a); ‘hosin (1447 Sek 9:29a,
13:35b; 1475 Nay 1:40a); 'ho'sitwo'ta (1481 Twusi 8:10b); 'hosi'na (1482 Kum-sam 5:10b); hosi n
i (1445 'Yong 107; 1447 Sek 6:9b, 23:53b; 1462 'Nung 2:92a; 1464 Kumkang 81b; 1482 Kum-sam
4:45a), 'ho'si.n i ’’la (1462 'Nung 2:49a, 1463 Pep 4:192-3); ho'sino'n i (1447 Sek 6:5b, 24:9a;
1459 Wei 1:10b, 1:25b); 'hosi'm ye (1459 Wei 2:58b), hosi'm ye ‘n (1449 Kok 36); hosilq (1449
Kok 50; 1447 Sek 23:52b, 53a); 'hosil 'ss oy (1445 'Yong 34, 92, 121; 1459 Wei 2:62b), 'hosil's oy
(1465 Wen 1:2:1:16b); ho'si.l i ’'la (1459 Wei 1:7-8); ho'si.l i 7 'ss oy (1445 'Yong 92, 1459 Wei
9:11b), ho'si.l i ’’la (1459 Wei 1:7-8); hosila ’n t oy (1481 Twusi 22:7b); hosin (1447 Sek 9:29a,
1459 Wei se:9a, 1475 Nay 1:40a), 'hosin 'ta "ma'ta (1459 Wei 1:15a); 'hosi'n i (1445 'Yong 42, 64;
1447 Sek 6:9b; 1459 Wei 1:52a; 1449 Kok 43; 1465 Wen 1:1:2:125b; 1482 Kum-sam 3:3b, 5:14a),
ho'si.n i ’ la (1447 Sek 23:42a, 1465 Wen se:6a, ? 1468- Mong 49b); ho'sino'ta (1447 Sek 13:26b);
ho'sinon (1447 Sek 6:5b), hosi'non (1447 Sek 23:22b); ho'sinon 'ka (1447 Sek 13:25b);
ho'sino'n i (1447 Sek 6:5b; 1459 Wei 1:10b, 2:69a, 9:11b, 9:35de; 1463 Pep 4:117a; 1465 Wen
se:6a); ho'sino'n i ' la (1465 Wen se:6a); ho'sino'n ywo (1447 Sek 13:26a); hosi'nwon (1462
'Nung 1:86a, 1463 Pep 5:169b), ho'sinwon t i (1459 Wei 17:42a); ho'si-ngi ’'ta (1459 Wei
21:218b), ho'sitwo'ta (1481 Twusi 8:10b);
84 PART I
A Reference Grammar of Korean
hosya (1447 Sek 6:4b, 6:9b, 6:17b, 13:27a; 1459 Wei 2:36b, 2:70b, 8:93b, 9:35de; 1462 'Nung
3:68b; 1463 Pep 2:231b; 1482 Kum-sam 2:2b), 'hosya (1459 Wei 8:93b, 1463 Pep 2:231b, 1475 Nay
2:1:43a, 1481 Twusi 22:46a), 'hosya za (1447 Sek 6:12a, 1463 Pep 1:16a); hosya'm i (1462 'Nung
10:42b, 1463 Pep 5:100a, 1482 Kum-sam 5:14a), ho'sya.m i'la (1459 Wei 14:58a), 'hosya'm ay
(1459 Wei 14:59a), 'hosya'm olwo (1482 Nam 1:33b), 'hosya'm on (1462 'Nung 3:2a, 4:13ab; 1463
Pep 6:145b, 7:180b); ho'sya'na (1465 Wen se:6a); ho'syan (1447 Sek 13:25b, 1463 Pep 6:144a),
hosyan (1447 Sek 6:7b, 1463 Pep 4:167a); ’ho'syal (1465 Wen 1:1:2:75b)
2.12.4.5. Bound stems.
It is difficult to predict the accent of later syllables in verb forms incorporating the common bound
stems showing status, respect, and aspect. We will assume a basic and etymological accent for -'k^a-
(and variants -G^a- and -'%-), the effective, and for the retrospective. That is needed to
account for such forms as:
retrospective nilk'ten (1481 Twusi 21:42b), nilo'ten (1459 Wei 9:36d), mwot'te'n i (1445 'Yong
9), is'ten t ay n’ (1464 Kumkang 79b), is'ten ta ( ? 1517- Pak 1:37b), is'tesin ka (1445 'Yong 88,
89 ), pat'te'n ywo ( ? 1517- Pak 1:19b), mek'te'la ( ? 1517- 'No 2:53b); ho'tan cyen'cho 'Iwo (1459 Wei
7:13b), ... .
effective cephu'kesi'n ywo (1449 Kok 123), kap'kan t i.n i ’’la (1459 Wei 18:18b), kos'ke'sin
ma'lon (1447 Sek 13:63a), me'ke'ta ( ? 1517- 'No 2:39a), me'ke'nul (1447 Sek 6:32a), is'kesi'ton
(1475 Nay 1:40a), na'ma is'kesi'nol (1447 Sek 23:56b), nilo'kesi'tun (1482 Kum-sam 4:50b) =
nilu'kesi'tun (1447 Sek 9:27a), nilo'kesi'na (1459 Wei 18:49b), talo'kesi'nul (1445 'Yong
101), 'ca'kesi'nol (1482 Nam 1:28b), 'wo'kesi'nol (1459 Wei 7:10a) = wo'nasi'nol (1447 Sek 6:44b),
'khu'kenul 'za (1482 Kum-sam 2:16a), ... .
Quite often the accent of the marker will be suppressed for prosodic or other reasons that are hard to
pinpoint. But no word of that sort will lack dots altogether: one or more of the other syllables will be
marked as accented.
The situation for the processive is more complicated, so we put its dot in parentheses in citing
the morpheme -(')no-. A number of the forms are attested as accentual doublets or near-doublets:
ho non (1459 Wei 21:215b; 1462 'Nung 1:77b, 8:124b; ?1517- 'No 1:35b; 1518 Sohak-cho 9:90a;
1588 Mayng 13:13a) but also honon (1447 Sek se:la, 9:33a; 1481 Twusi 7:1b); 'ho'no.n i (1586
Sohak 2:30b, 4:43a) but [‘be many’] hano'n i (1445 'Yong 2, Manlyek text; 1459 Wei l:30ab; 1482
Nam 1:36b); 'ho'no.n i ’ la (1447 Sek 6:5b, 13:2a; 1459 Wei 1:11a, 1:23b, 2:2a; 1462 'Nung 6:43a, 8:
86 b; 1481 Twusi20:34b; ? 1517- Pak 1:58a; 1586 Sohak2:25a) but also hono n i ’la (1482 Nam 2:6ab)
and hono.n i ' la (1481 Twusi 16:91b).
kanon (1447 Sek 6:19a, 1481 Twusi 7:10b) and kano'n i (1445 'Yong 2, 1447 Sek 6:9b) but
ka'non 'ta ( ? 1517- 'No 1:1a)
ho'nwon (1459 Wei 13:35b, 1462 'Nung 1:90b) but also honwon (1447 Sek se:6a)
hosi 'non (1447 Sek 23:22b) but also ho 'sinon (1447 Sek 6:5b)
"a'no.n i ’ la (1482 Kum-sam 2:3a) and pwo'no.n i ’’la (1459 Wei 21:206a, 1465 Wen 1:2:1:39b)
but also "sano'n i (1447 Sek 13:10a, 1481 Twusi 25:23a)
Processive modifier forms are usually unaccented: ■■■[']non. The common form is.non - ‘ ••• that
is/stays’ is always unaccented, and that is true also of its contracted versions ’ys.non and ’s.non with
rare exceptions that imply *is'non as the model: towoy'ye ’ys'no'n i (1481 Twusi 8:42a, ~ 'Gwo
7:26b), towoy'ye ’ys'no'n ywo (1481 Twusi 8:42a); maf'Jka ’s'non ( ? 1517~ Pak 1:40a).
Processive indicative assertive forms (■■• no'ta ) never carry the accent on the aspect marker:
hono'ta (1447 Sek 6:2a, 6:14b; 1481 Twusi 7:2a, 8:1b, 8:52a), "ano'ta (1462 'Nung 2:114b),
i'kuyno'ta (1481 Twusi 15:6a), mwo'lono'ta (1462 'Nung 1:16b), mekno'ta (1481 Twusi 25:18a).
When the modulator -'w u /o- is attached to the processive, the combined form -'nwo- (sometimes
-'nwu-) is usually accented: ho'nwon (1459 Wei 13:35b, 1462 'Nung 1:90b), is'nwo'n i (1447 Sek
6:20a), tut'nwon ka (1449 Kok 2), towoy'nwo'n i (1463 Pep 2:28b), cap'nwola (1481 Twusi 10:7b),
ip'nwola (1481 Twusi 8:42a), hosi'nwon (1462 'Nung 1:86a), wolm'ki'si'nwo'swo-ngi ’’ta (1463 Pep
A Reference Grammar of Korean
PART I 85
2:47a). But not in these examples: 'honwon (1447 Sek se:6a), [’Ngwen] honwon t un ... (1579 Kwikam
1:24b); 'honwo'la (1459 Wei 10:4b); "mwutnwo'la (1481 Twusi 22:39b) = "mwunnwo'la (1481
Twusi 16:39b); po'lanwo'la (1447 Sek se:6a); wonwos'ta (1481 Twusi 7:39a); 'twu-’ys.nwon (1482
Nam 1:15a); nilu'sinwon (1447 Sek 9:35b; Cf 1459 Wei 55-6).
After the honorific marker, the processive is sometimes accented - hosi'non (1447 Sek 23:22b),
but usually not: ho'sinon (1447 Sek 6:5b), ho'sino'ta (1447 Sek 13:26b); sisu'sinon 'ka (1449 Kok
124), tas.ko'sino'n i (1447 Sek 6:12a), nilu'sino'n i (1447 Sek 13:47b, 1462 'Nung 1:38a), ka'sinon
(1459 Wei 2:52a), 'nye'sino'n i ’-ngi s 'kwo (1447 Sek 6:23a), "alo'sino'n i ’-ngi s ka (1447 Sek
6:14-5). After the deferential - "zop- the processive morpheme is normally not accented, but there are a
couple of exceptions: cwo"ccop'non (1463 Pep 1:24b), ki'tuli'zop'no.n i ’’la (1447 Sek 24:5b).
When the polite marker -ngi is attached the processive morpheme is always accented (-'no-ngi ■■■):
pho'no-ngi ’ta (1459 Wei 8:94b), 'wo'no-ngi ’’ta (1447 Sek 6:29b), wosi'no-ngi ’ta (1459 Wei
10:8a). And -ngi itself never carries an accent.
The underlying accent of the honorific marker -%'si- often surfaces: ho'sino n (1459 Wei 1:
25b), ho'sino n i (1447 Sek 6:5b; 1459 Wei 1:10b, 2:69a), ho'sino n i ’’la (1465 Wen se:6a),
ho'sino n i ’-ngi s 'kwo (1447 Sek 24:9a, 1459 Wei 9:35de), ho'sino n i 'si n i ’’la (1463 Pep
4:117a); 'ho sino 'ta (1447 Sek 13:26b), 'ho'siten 'ka ( ? 1517- Pak 1:51a); kwutu'si'ta (1463 Pep
2:173a); nwopho'si'kwo (1463 Pep 2:173a); tulu'si'kwo (1447 Sek 13:30b); sisu'sinon 'ka (1449 Kok
124); nilku'si'nwon (1465 Wen se:68a); ceho'sya (1449 Kok 46), cwocho'sya (1459 Wei 8:93b, 1463
Pep 3:19b), meku'sya (1459 Wei 10:9a), anco'sya (1459 Wei 8:101a); ... . Sometimes the accent
appears on the preceding vowel: "ep'susi'kwo (1462 'Nung 1:18b), "ep'susi'ta (1449 Kok 53),
"ep'susya'm i (1463 Pep 2:15-6) - Cf "ep.su'sya (1463 Pep 2:22a); "sa'mosi'n i (1447 Sek 6:4a);
"a'losi'm ye (1465 Wen 1:2:3:6a); te'pu'lusi'n i (1449 Kok 52); "wu'lusi'kwo (1459 Wei 8:101a);
'tu'lusya'm i (1447 Sek 23:44a); twuthe'wusi'm ye (1462 'Nung 10:42b); ... . We can regard this as a
prosodic displacement which pushes the accent back a syllable; there are no examples of the dot
appearing on both the sibilant syllable and the epenthetic vowel. When there is a dot on the syllable
preceding - si- it is usually part of the accent pattern of the stem: mwo'lo'sya (1445 'Yong 19, 1459
Wei 21:210b), na'thwo'sya.l i ’’la (1459 Wei 17:78b), il'Gwu'syan (1459 Wei 21:218b), tho'si'l i
’le'la ( ? 1517- Pak 1:64b), ne'ki'sya (1447 Sek 6:17b), a'ni ’’sya (1463 Pep 2:6a).
The basic rising accent of the deferential -"zoW- is the result of contraction from *-zo'po- (<
*-oso'po-) and it surfaces for some of the forms: cap 'sopke'n i (1459 Wei 21:203a), mak "sopke'nul
(1459 Wei 10:1b), ilkhot"copnwon (1482 Kum-sam 4:11b), pat"copte'la (1459 Wei 2:37b), tut'copkwo
'za (1449 Kok 106).But in many of the forms only the low pitch survives: hozo'Wa (1447 Sek
24:5b), "pwozo'Wa.n i (1459 Wei 8:17b), 'kazo'Wwon (1459 Wei 8:92b), tutco'Wolq (1447 Sek 9:2a),
"a'zoWol 'kka (1445 'Yong 43), 'hozop'kwo (1447 Sek 6:1b, 24:5b), "sitcop'key (1447 Sek 24:10b),
'pwozop'ta (1459 Wei 8:28a, 18:81a), hozop ten (1447 Sek 13:51a).Sometimes, instead, a high
pitch appears, as the result of a prosodic displacement from the following element: 'pwo'zowa 'two
(1462 'Nung 1:47a), 'ka'zoWa 'za (1447 Sek 23:40a), 'ho'zoWo'm ye (1447 Sek 6:17a), ho'zopke'na
(1447 Sek 13:53b), ho'zopno'n i (1463 Pep 5:186a), ... .
2.12.5. Accent and spelling in Middle Korean texts.
The accent dots of Middle Korean were written to the left of the syllable and therefore vary in
where they stand in the stream of phonemes depending on the extent to which morpheme identifications
are permitted to override phonetic considerations. The text Wei.in chen-kang ci kok (= 1449 Kok)
normally separates particles from a preceding noun that ends in a resonant: ye'lum ul (99), 'nom i...
nom ol (11), ema "nim i (17), mozom o'lan (121); nwun 'ey (2), cey 'kan ol (40); sal i (41),
'stol ol ... mye'nol i (36); ... . But forms of the substantive are excepted: pwus'kulywo'm i (120). And
the syllabification is phonetic for nouns with a voiceless final: ti p ul (45); pa p ol (122); kwo'c i (7),
mi'th uy non ... (70). Noun + copula, like verb stem + ending, was left unanalyzed, and that accounts
for the syllabification of Ikwo-mill in 'mwom i ••• ceyye'kwo.m il ss oy (134).
Sekpo sangcel (= 1447 Sek) spells noun + particle according to the spoken syllables, but certain
nouns ending in -ng are excepted, probably because they are clearly of Chinese origin: cywungsoyng
86 PART I
A Reference Grammar of Korean
i (6:19b), "yang o'lwo (6:24b); Cf "cywu'ng uy (6:19a), susu'ng uy (6:29b), for which the Chinese
origin is less apparent. Examples of the usual syllabification: no'm oy (se:6a); kul ul (se:6a); pu l i
(6:33a); mwu'l i (6:28a); sto'l i (6:13b, 14a), sto'l ol (6:15a), a to 7 i (6:5a, 9b), a'to'l ol (6:3b,
5b), a'to'l oy (13:19a); ha'no'l i (6:35b), ha'no'l on (6:36a) "i'l ol (6:8a, 18a, 26a, 27a; 9:5a; 13:
33b; 19:40b), "i'l i (6:9a; 9:17b; 19:10b, 20b, 24b) "i'l oy (se:5b); "ma'l ol (6:8b, 13:47b, 24:1b),
"ma'l on (6:36a, 9:27a), "ma 'l i (6:36a, 25:53b) tuthu'l i (6:30b); ci 'p uy 'sye (6:16a), tul cci 'p i (6:
35b); "hoyngtye'k ul (1447 Sek 6:2b); nye'k ul (6:25a), nye'k uy sye (6:33b), nye'k o'lwo n’ (1446
Sek 6:3a); ce k i... (6:40a), ce k uy (6:19a), ce 'k u 'lan (6:1 la).
The text of 'Yongpi echen ka (= 1445 'Yong) follows the spoken syllables: no m i (48), no m on
(77); 'kwulwu'm i (42), mozo'm ol (85), "nim-'ku'm i (49) hyen pe n ul (31); mo l i (31), nwun z
mu'l ul (91), "mil mu'l i (67), ku'l ul (7), "pye'l i (50, 101), palo'l ay (2); twoco'k i (33), twoco'k
ol (19, 115); kwo t ol (110), 'kwo 't ay (26); ptu t i (8); na 'c oy (101).
The spellings of Nayhwun (= 1475 Nay) are similar, but exceptions are made for certain
morphemes with final resonants, some of which are of obvious Chinese origin ( kesang ul 1:70b,
"cams.kan ina 2:1:2b, si'cyel 'ey 3:32a, chapan ul 2:2:73b, kanan i 2:2:59b) though others are
not: "nim-kum i (10a), swu'l ul (3:61a). Cf mwo.m o'lwo (2:1:30a), no'm oy (1:9a), mozo'm ay
(se:6a); 'swo'n i (1:18a); "ma.l i (2:2:47b), he'mu.l i (1:84a), "i.l ol (1:53a, 1:84a, 2:1:40b).
Twusi enhay (= 1481 Twusi) follows the pronunciation in syllabifying noun + particle, but
makes a few exceptions for morphemes with final resonants, both Chinese (" sya'wong ol 25:9b) and
non-Chinese ( swul ol 15:38a, yet for the same word also swu'l ul 8:34a).
The later version of Sohak enhay (= 1586 Sohak) generally demarcated the particles: "kye'sim
'ay (6:122a) = "kye'sim 'ay (See p. 267), ancum ul (3:9b), a'chom uy (4:33a), kunsim ul (4:9b);
"pyeng i (1586 Sohak 6:27a), "syeng i (8:37b); kul ul (6:102b), "mal i (5:95b); swon 'ay choyk
ul (6:102b); ... . There were a few lexicalized exceptions: cwuk'u.m wolwo (2:11a). The adverb
"man'il (2:4b) ‘if’ reflects the Chinese source ‘one in ten thousand [chances]’. And 1586 Sohak
overanalyzed "nwul ul (6:58a) = "nwu lul (1449 Kok 52) ‘whom’. The earlier version known as
Pen.yek Sohak enhay (= 1518 Sohak-cho) syllabified phonetically: ka'zo'mye'lwo 'm on (9:90a) - Cf
ka'omyel'wom un (1586 Sohak 6:83b); "sa'lo'm uy (8:22a) - Cf "sa'lom i (5:48b); no'm oy
(8:15a); il'hwu m i (8:2b); swo'n i (10:3a) - Cf 'swo'n ol (6:102b). Occasionally it conflated -l ul
to lul, as in na 'lafh] ta 'soliten il lul [< "il ul] (9:39a) = ta 'solim ul (1586 Sohak 6:35b).
None of the texts had a way to write final h or consonant clusters, so they were always syllabified
phonetically: "nay'h i (2); sta'h ay (1449 Kok 41), 'sta'h ol (1447 Sek 6:19a); wu'h u'lwo (1447 Sek
13:13b); tu'lu'h ey (1445 'Yong 69); "twuy.h ey nun ... al'ph oy non ... (1445 'Yong 30); kil'h ul
(1447 Sek 6:19a, 1449 Kok 86); ha 'nol'h i (1445 'Yong 21, 30, 34); 'tol'h i (1449 Kok 11); an h ay
(1475 Nay 1:4b); nam'k ol (1449 Kok 86), nam'k i (1449 Kok 99); nyen'k ul (1445 'Yong 20), nyen'k
i (1447 Sek 6:22b, 24:43b); twos'k ol (1475 Nay 10a); hon "na'th ay (1445 'Yong 47).
3.0. Words.
The description of Korean grammar in this book is based on a division of Korean sentences into
WORDS, and an assignment of each word to a PART OF speech. The decisions on word boundaries are
based on syntactic criteria, and therefore they are more generous than the decisions that underlie the
writing of spaces in Hankul texts, where the criterion is purely phonological, based on the likelihood
of pause when a sentence is said. A short word, such as a postpositional particle that marks the
grammatical function of a noun phrase, is usually joined to the preceding word as if a suffix, so that
you will not hear an overt pause or slowdown between the words; but a silent grammatical juncture
lurks just below the surface of the structure, and we find it convenient to reveal that with a space in
our Romanized sentences.
3.1. Inflected and uninflected words.
On the basis of internal structure, the words of Korean clearly fall into two classes: inflected and
uninflected. Each inflected word consists of a stem + an ending. The stem (sometimes called the
A Reference Grammar of Korean
PART I 87
BASE) belongs to a large but limited class of constituents which do not occur except with the
attachment of one of a much smaller class of endings; the endings do not occur except when attached to
a stem. Apparent exceptions:
(1) Derivationally related nouns and verb stems:
hemul < he'mul ‘error, misdeed’, hemu-l- (< ? ) ‘tear down’
II < "il ‘event, happening, matter; job, work’, "il- ‘come into being, happen’
kem ‘black checker’ (= hukci), kem- < "kem- ‘be black’
kil < "kil ‘fathom’, kl-l- < "kil-‘ be long’
kot < "kwot ‘straightway’, kot.ta < kwot- ‘be straight’ (or 'kwot ‘place’?)
kkwumi ‘beef shreds’, kkwumi- < 'skwu'mi- ‘decorate’
mak < mak ‘last’, mak- < mak- ‘block, obstruct; ... ; complete, put an end to, ... ’
phum < *'phwum (-* phwu'm ey) ‘bosom’, phum- < phwum- ‘carry in the bosom’
pis < pis ‘comb’, pis- < pis- ‘comb the hair’
pophul ‘nap’, pophu-l- ‘(cloth) has a nap’
ppyem ‘span’, ppyem- ‘measure by the span’
sin < 'sin ‘shoe’, sin- < "sin- ‘wear on the feet’
sol ‘skin pustules’, so-l- ‘be itchy and sore’
tel ‘less’, te-1- < "tel- ‘lessen’
tti < 'stuy ‘belt’, tti- < stuy- ‘wear (a belt)’
In the case of -L- stems we might conclude that the noun is the imperfect adnominal (= prospective
modifier), deriving kil ‘fathom’ from kllq ‘to be long’, but there is little to argue against treating such
cases as simply the stem. The noun of the pair kamul < "ko'mol ‘drought’ and kamu-1- < "ko'mol-
‘go without rain’ could be a reduction of kamulm, the regular substantive of the verb, in contrast with
the irregular derived substantive kamul.um ‘drought’, which preserves the expected earlier form of the
substantive in its uncontracted form (kamulm ‘going without rain’ < *'ko'mo'lom ); unfortunately, we
lack attestations until around 1700 for either the verb stem or the noun. Neither as ‘year of age’ nor as
‘flesh’ can sal be directly related to sa-I- < "sal- ‘live’ for the nouns earlier had the low-back vowel
/o/, MK sol and "solh, respectively. Similar: nal < nolh ‘warp’, na-l- < nol- ‘thread the warp (of a
loom)’. Nor can we easily relate an ‘inside’ to an- < "an- ‘clasp to one’s bosom’ because the noun
was earlier 'anh (as attested by the h preserved in anphakk < anh pakk ‘inside and outside’), nor els
‘gesture’ to ci(s)- < "ciz- ‘make, do’ because the noun apparently comes from an earlier "cus
‘appearance’.
We should consider also those cases of derived nouns and adverbs that coincide with a stem having
final -i- or -y-, such as kalkhwi ‘a rake’ and kalkhwi- ‘to rake’, toy < 'twoy ‘a measure’ and toy-
< 'twoy- ‘to measure’, kkoy < 'skwoy ‘ruse’ and kkoy- < "skwoy- ‘cheat out of’. When verb stem
and noun coincide in shape it is hard to decide which came first; often all we can say is that the two
are derivationally related. In some instances the meaning of one of the pair is clearly secondary: ai lul
pay- < poy- ‘conceive a child’ must come from pay < "poy ‘belly’, not the other way around.
(2) The last word of Salam sallye cwu ‘Save me!’ and Na com cwu ‘Give me some!’ is a
contraction of cwuu < cwuo; compare cwuso contracted (by way of cwusyo) from cwusio ‘give!’.
Similar are I ke(s) po ‘Look at this!’ and Ka-po ‘Go and see!’ < poo. CM 1:119 is confused about
these forms and those of (4) below.
(3) The stems i- ‘it is’ and ha- ‘does; is’ are often abbreviated to zero, leaving the endings standing
as if free. In this book the abbreviation is shown by an apostrophe: twu si ’myen ‘if it’s two o’clock’,
pata ’ci ‘it’s the sea’; ka ya ’keyss.ta ‘I’ll have to go’.
(4) The infinitive (§9.4) has a zero alternant after certain stems, e.g. ka from ka- ‘go’. In the
intimate style we find commands like Ese ka ‘Go on!’ and I ke 1’ sa ‘Buy this!’
(5) There are a few odd abbreviations like po’ to tut’ to mos hanta = poci to tut.ci to mbs hanta
‘can neither see nor hear’ (similarly o’ to ka’ to mbs hanta ‘can neither come nor go’, olu’ to nayli’
to mbs hanta ‘can neither rise nor fall’), and ...-’ tus = ...-nun tus ‘seemingly ( ...-ing)’. In Middle
88 PART I
A Reference Grammar of Korean
Korean the t- forms could attach directly to the stem, and stems elided the liquid. Starchevskiy
(1890:668) described that and included the 'sa version of the particle 'za, the source of (i)ya, as well
as 'two and 'tus\ there are MK examples of that in Part II.
Unlike stems, uninflected words occur freely without the requirement that something be attached.
There is a class of particles and they are very similar to the verb endings in some respects, but the
nouns occur freely without a particle, in a great variety of environments, and many of the particles are
found attached to verb forms (stem + ending) as well as to nouns.
3.2. Parts of speech.
In this book all the inflected words are called verbs. There are, to be sure, many subclasses, but
they share the characteristic of being stems that require the attachment of one of the inflectional
endings in order to serve as a free word. The uninflected words divide into two broad categories called
nouns and PARTICLES. The characteristic of particles, which are typically quite short, is that they
occur as the last member of a phrase, or as part of a string that can be treated as the last member. In
pronunciation they are attached to the last word of the phrase: a noun or a verb form or some other
particle. These characteristics they share with forms of the copula (i-), a secondary subclass of verb.
Typically, particles occur after nouns, but some of them also are attached to verb forms, not only those
forms which often serve as the head of a nominal phrase (as does haki ‘doing’) or an adverbial phrase
(as does hakey ‘so as to do’) but also forms which often stand as the head of a verbal phrase, such as
hay ‘does’. Particles that are case markers specify valences that certain other languages express
through affixes or prepositions; contextual particles (delimiters and particles of focus) convey
information carried in many other languages by articles, adverbs, prosodic elements, or word order.
Many of the particles originated as bound nouns or verb forms.
The verb ending is considered to be in construction only with its stem. The particle, on the other
hand, is taken to be in construction with the entire phrase, which may end in a noun, a verb form, or
one or more prior-attached particles. In a string of particles, the constituents are assumed to peel off
from the right, even when two (such as ey se) form a frequent collocation.
3.3. Free and bound words.
The borderline between “free” and “bound” forms is not always easy to delineate and many
decisions have to be somewhat arbitrary, but I try to be as consistent as possible. Every word is to
some extent “bound” in that there are constraints upon its occurrence: ‘eats’ can take as its object
‘rice’ but not ‘his high-jumping’ though ‘likes’ can take both phrases as objects. But some words are
much more severely bound than others, so that it is easier to list their constraints in general terms. In
this book I speak of FREE and quasi-free nouns, of FREE and bound verbs. I also speak of certain
words as being separable or inseparable, meaning that elements (such as particles) either can or
cannot be inserted between those words and the other words with which they are typically in
construction. Another kind of word category is that of the bound nouns, nouns that are restricted in
construction to limited sets of partners. The bound (“prenoun” =) adnoun approaches the status of a
noun prefix, the bound “postnoun” approaches that of a noun suffix. The bound “preverb” approaches
the status of a verb prefix or an adverb, the bound “postverb” approaches that of a verb suffix.
Ultimately perhaps all noun prefixes should be called “bound adnouns” and all noun suffixes should be
called “bound postnouns”, but I have made a division, based on the relative range of occurrence with
different sets of nouns.
In the same sort of way, it might be said that those particles which appear only after nominals
approach the noun subcategory of “postnouns” and those which appear only after verbals approach the
category of verb endings. The difference lies in the constituency of the phrases: I consider that the
particle always stands in construction with the entire remainder of the phrase.
3.4. Ionized parts of speech.
The chart labeled parts of speech (pp 90-91) presents a detailed overview of my analysis of
Korean word types. We find three “polar” categories: the major groups of noun and verb, the minor
A Reference Grammar of Korean
PART I 89
group particle. There are numerous subcategories, some interrelated in complex ways that I have
tried to capture in the chart. The categories of verbal noun and postnominal verb are “ionized” in
that they show what seem like chemical bonds (attractions) between the categories of noun and verb.
The verbal nouns occur in phrases that are in construction with postnominal verbs. Some of the verbal
nouns also turn up in other constructions as ordinary nouns, and some of the postnominal verbs also
occur as other subcategories of verb. Certain verbal nouns are separable, in that they can form a
phrase with additional elements (such as particles) before the appearance of the postnominal verb,
while other verbal nouns are inseparable. And some of the postnominal verbs are separable, in that a
particle or the like can intervene between the preceding noun and the postnominal verb, while others
are inseparable. Frequently an inseparable verbal noun turns out be also PRE-INSEPARABLE (that is, it
combines only with inseparable postnominal verbs), but some are pre-separable for they occur
with postnominal verbs that, with certain other verbal nouns, are separable. Auxiliary verbs are also
divided into separable and inseparable.
The class of verbs divides into transitive and intransitive, and the intransitive includes the
subcategory of adjective (= descriptive verb), which in turn has the one-member subcategory of
copula, a general noun-predicator. In the same sort of way, verbal nouns have properties of transitive
(vnt = transitive verbal noun), intransitive (vni = intransitive verbal noun), and adjectival (adj-n =
adjectival noun). There are a few verbal nouns that are defective, for they occur with only a few
paradigmatic forms of the postnominal verb; there are also a few defective verbs. In earlier
treatments I have restricted the terms “verbal noun” and “adjectival noun” to constructions with the
postnominal verb and adjective ha- ‘do/be’, since those are the most common. But the extended
treatment offered here is logically more consistent. A list of examples of each of the part-of-speech
categories will be found in later sections (§§5-7), and information on the constructions involved should
be sought in those sections and in the appropriate entries of Part II.
3.5. Shortened words.
Some words frequently appear in shortened form. We have already (§3.1) called attention to the
“zero” abbreviation of ha- ‘do/say/be’ and i- ‘be’. After a vowel these stems may fail to emerge and
that leaves just the endings, standing alone as if free words: -V ’myen = hamyen ‘if it does/is’ or =
imyen ‘if it is’. The suppression of i- takes place after a vowel unless the stem is itself followed by a
vowel (as in the past iess.ta or the infinitive ie), when i- merely reduces to y- after a vowel: kama ’ta
‘it’s the oven’ but kama yess.ta ‘it was the oven’. The suppression of i- is so common after a vowel as
to be considered standard (an alternant of the copula rather than a variant), but the abbreviation of ha-
is somewhat less predictable and therefore, except in a few complex endings like -ulye ’myen, it is
usually treated as a shortened variant. The difference in the way that suppressed ha- and suppressed i-
are treated reflects the fact that the phonological bondage of the copula with a preceding word is closer
than that of ha-. In ka ya ’keyss.ta ‘I’ll have to go’ there is no pause, but one can be inserted with the
ha- restored as ka ya I hakeyss.ta ‘I will have to go’. On the other hand, it sounds pedantic or
bookish to say wuli nala ita for wuli nala ’ta ‘it’s our country’.
The shortened forms of the plain and prospective modifiers of these two verbs (’n = han ‘that
did; that is’ or = in ‘that is’, and ’l(q) = hal(q) ‘that is to do/say/be’ or = il(q) ‘that is to be’) are
homonyms with the shortened forms of the postvocalic shape of the topic and object particles (n’ =
nun, 1’ = lul). In our Romanization we distinguish them by the location of the apostrophe: uysa ’n
(salam) ‘(a person) who is a doctor’, uysa n’ ‘as for the doctor’. The “zero” abbreviation of the
processive modifier of ha- (’nun = hanun) is distinguished only by our apostrophe from the
homophonous full postvocalic shape of the topic particle nun: cangsa ’nun (salam) ‘(a person) who is
engaging in business’, cangsa nun ‘as for business’. When writing, Koreans do well to avoid
abbreviations as much as possible. It makes the content easier to understand if you write out in full the
forms of the copula and ha- wherever they occur in a Hankul text, and spell particles in their full
forms, as well as taking care to specify the grammatical role of phrases by marking them with the
appropriate particles more often than is usual in speech, where case markers are casually dropped.
90 PART I
A Reference Grammar of Korean
Chart: Parts of speech (left)
bnd
postn (> suffix)
bnd
pre-
pel
bnd
pren
--BOUND.
\
\
\
\
bnd preverb
(> adv,
prefix)
(> prefix)
postsubst
/ \
_1
1 \
_2 ..
postn count
.
quasi
-free .-.
.
\ 1 \
•
\ | \ postmod
.
\ 1 .•
3-.
\| / \
•
NOUNS -
----FREE \
4.
adv
/
\ \ \
.
.-
--/
\ \ \
•
1
num
\ proper \
1
\ noun 6_
_5 .
interj
\
\..vni-
deictic
1
pren noun
1
Verbal Nouns
/ I
/ I
/ vnt
. *-sep
/
/
/
/
T /
. *-insep /
adj-n | /
I /
I /
_I /
I
|_pre-insep
/
defective
I_pre-sep
quasi-pcl n forms
\ I /
PARTICLES <.
/ I \
(postn) <.onliy after noun | onlly after verb-> (v ending)
1 postsubst adj-n insep: cik
2 postnom/postmod adj-n insep: man
3 postmod adj-n sep: pep, tus
4 postmod adj-n insep: ak, man, (p)pen, mu
5 postmod vni insep: chek, (s)sa
6 pre-insep postmod: seng/sang
A Reference Grammar of Korean
PART I 91
Chart: Parts of speech (right)
bound postverbs
(> suffixes) defective
1
infinitives
1
/
1
/
1
•defective
1
/
/
1
/
1
/
Stems -* VERBS <- Endings
1
i
1
vi---..
.vt
pseudo-vi /
/ -
-\
insep / .
qvi
\ insep
\ / 1
(iss-, eps-,
\ /
Postnominal
-ess-, -keyss-)
aux
/ \ 1
/ \
sep \ |
/ sep
ADJ
-/
l /
COP
<— quasi-pci v forms
92 PART I
A Reference Grammar of Korean
There are various other sorts of common abbreviations:
(1) ke = kes ‘thing, one, fact’; mue, mwe, mwes = mues ‘what, something’; an - = ani ‘not’; key
= keki ‘there’ or = (ku) kes i ‘(that) one’; cey = ceki ‘there’; yey = yeki ‘here’; -ulq cey = -ulq
cek (ey) ‘(at) the time that’; anh- = an ha- = ani ha- ‘do/say/be not’; -canh- = had anh- (= haci
ani ha-) ‘do/say/be not’; mas i ’ss.ta (mas-iss.ta) = mas i iss.ta ‘is tasty’; caym’ iss.ta or caymi
’ss.ta. Notice that /masitta/ is mas i ’ss.ta with the particle overtly present because the synonymous
phrase mas iss.ta (without the particle i) is /matitta/.
(2) il’ lo = ili lo ‘(toward) this way’; il’ one la, ili on’, il’ on’ = ili one la ‘come here!’; k’an twu-
= ku man twu- ‘leave it alone’; et’ ’ta = eti (ey) ’ta ‘to where’; kac’ ’ta cwu- = kacye ’ta cwu- =
kade ’ta cwu- ‘bring it (for someone)’; teykko o- < teyliko o- ‘bring him along’.
(3) The nominative particle i is shortened to y (the front component combining with a preceding
vowel) in some of its irregular appearances after a vowel, where we would expect the suppletive
alternant ka, and the copula i- is sometimes reduced to y in a similar way: (kes -* ke -*) key = kes i
or kes i-. The nominative forms nay ‘I’, cey ‘I [formal]’, and ney ‘you’ appear either alone or
pleonastically followed by the particle ka; they are historically from na, ce, and ne + this reduction of
the particle i to y. The identical-sounding nay ‘my’, cey ‘my [formal]’, and ney ‘your’, on the other
hand, are contractions of na uy ‘of me’, ce uy ‘of me [formal]’, and ne uy ‘of you’. The pleonastic
sequence i ka is used for i in the northeast, in Hamkyeng, according to Mkk 1960:4:26, which has the
examples chayk i ka iss.ta = chayk i iss.ta ‘there is a book’ and sensayng i ka osinta = sensayng i
osinta ‘the teacher comes here’. This usage has been reported not only in dialects of Hamkyeng but
also of Kyengsang.
(4) After most adjectival nouns, forms of ha- ‘do/be’ which consist of the stem with an attached
ending that begins with t, c, or k have shortened variants, in which the vowel drops and h undergoes
metathesis (that is, it switches positions) with the voiceless consonant: tha for hata, chi for haci, kho
for hako. Examples are kantan tha ‘is simple’, kantan kho ‘being simple’, and kantan chi ‘is simple,
I suppose’. After certain processive verbal nouns, a few forms of ha- shorten in the same way:
sayngkak hakey -* sayngkak khey ( -» sayngkakhey, Cf §2.6) ‘so that one thinks’. The shortened
version of the suspective haci ‘does’ appears only in negative expressions, where it is optional. The
variant has two shapes: chi after a typically voiced sound (as in salang chi anh.nunta ‘does not love’)
and ’ci after a typically voiceless sound (as in sayngkak ’ci anh.nunta ‘does not think’). After an
adjectival noun the shortening of haci ‘is’ is usually chi regardless of the preceding sound, as in
nek.nek chi anh.ta ‘is not enough’ and phyen.an chi anh.ta ‘is not comfortable’. But many people
seem to use ’ci and chi in free variation with both descriptive and processive verbal nouns. In the 15th
century the short variants enjoyed wider use, including examples of ‘do’ as a transitive verb: C1N ol
khwo 'eye ho rn ye n’ (1462 *Nung 7:73b) ‘if one wants to do the true thing’. The aspirated forms
( tha < ho ta, thi < ho 'ti, 'khwo < ho 'kwo, 'khey < ho key / khuy < ho 'kuy) occur only after
voiced sounds, and the unaspirated forms {’ ta, ’ ti, ’ kwo, ’ key, ’ kuy, ... ) appear only after voiceless
sounds, but the -LQ of Chinese loanwords, though probably pronounced just as III, was sometimes
treated as voiceless (See p. 50): pat non 'hwoki lol kywelq ’key ho'n i (1447 Sek 9:19-20) ‘[he] let
them divide the fields’. (The Middle Chinese source of -LQ was an unreleased final It/.)
(5) There are a few examples of dropped p (Cf §2.11.2): phul-’ath [dialect] = phul path ‘weedy
spot, bush, thicket’; si’-wel *- sip-wel ‘October’ (we would expect sip-’el, see §2.7.4) and si’-o li =
sip-o li ‘15 leagues’; ka’-o [dialect?] kap-o (a cyclical binom); ttelum ha- (< ttelp- + -um) ‘be a bit
astringent’. There are also a few examples of dropped k, notably 'yu’-wel <- 'yuk-wel ‘June’ (but this
may have been influenced by the poetic name for June tyiwel ‘flowing month’) and onyu’-wel ‘May or
June’. But mokwa < "mwo-'kwa (1527 Cahoy 1:6a= 11a) = mok.kwa < mwok-kwa ‘Chinese
quince’ and Paychen < Paykchen (placename) are variant borrowings. The texts provide at least one
example of m dropped after p\ 'pap-e'ki ( ? 1517* 'No 1:45b; the initial circle is too round to be an m )
< pap me'ki (1481 Twusi 15:4b) ‘eating, having one’s meal’.
A Reference Grammar of Korean
PART I 93
(6) Final 1 drops in many words when they serve as the first member of a compound that has the next
member beginning with an apical articulation:
kyewul ‘winter’, kyewu’ sal.i ‘winter garb’, kyewu’ nay (also kyewul nay with /ll/) ‘all through
the winter’
kaul ‘autumn’, kau’ nay ‘all through the autumn’
hanul ‘heaven’, hanu’ nim ‘God’
pul ‘fire’, pu’ napi ‘moth ( *- fire butterfly)’, pu’-nemki ‘a kind of stove’, pu’-ce ‘fire
tongs (•*- chopsticks)’, pu’ son ‘fire scoop ( *- hand)’, pu’ sap ‘fire shovel’
atul ‘son’, atu’ nim ‘esteemed son’
ttal ‘daughter’, tta’ nim ‘esteemed daughter’
chal ‘sticky’, cha’ co ‘glutinous millet’, cha’ tol ‘flint ( *- sticky stone)’, cha’ pssal
(spelled chap-ssal) ‘glutinous rice’
hwal ‘bow’, hwa’ sal ‘arrow’
mal ‘peck’, ma’-toy ‘pecks and measurefuls’
mul ‘water’, mu’-cawi ‘pump’, mu’-nem.i ‘(?) overflow’, mu’ com ‘athlete’s foot, fungus’
mul ‘dye’, mu’-sayk ‘dyed color’
panul ‘needle’, panu’ cil ‘needlework’, pan’-cit koli ‘sewing box’ < *panu’-ci[l] s kwo'li
kumul ‘net’, kumu’ cil ‘netting’
kempul ‘dried grass and twigs’, kampu’ namu ‘tinder’
petul ‘willow’, petu’ namu ‘willow tree’
sol ‘pine’, so’ namu (also sol namu) ‘pinetree’
ssal ‘rice’, ssa’-cen ‘rice store’
ipul ‘bedclothes’, ipu’ cali ‘bedclothes and mattress’
cwul ‘line’, cwu’ tay ‘fishing line and pole’
nal ‘day’, na’-nal-i ‘day by day’
tal ‘month’, ta’-tal-i ‘month by month’
si-nay ‘streamlet’ (from sil ‘thread, ... ’ 4- nay ‘stream’)
si’-tha(y) ‘ox pack’ (from sit-/sil- ‘load’ + thay/tha = cim ‘burden’)
Notice also kwutu-ccil ‘hypocaust work’ and kwutwu-ttay ‘hypocaust’ from kwutwul ‘hypocaust’ (by
by way of *kwutwu[lj s cil, *kwutwu[lj s tayl)\ pha’-il from phal-il ‘8 April = Buddha’s birthday’.
This dropping of the liquid before an apical consonant was a general phenomenon in Middle Korean
(Cf §2.11.3), and it regularly occurred in verb forms such as "ma 'ti and "ma 'ta for modern malci and
malta. The modern language retains the basic -1- of such stems except before n (manun) and s
(masinta), or in fossilized forms such as -ca maca. The elision of 1 before n is no longer productive,
however, and newer formations regularly have l.n, pronounced /II/. See nim in Part II. Some dialects
apparently never elided the 1, e.g. that of Ceycwu (LSN 1978:18): toltol-i ‘monthly’, nalnal-i =
Inallalil ‘daily’, ttol-nim = Ittolliml ‘your daughter’. On the other hand, the 'Yukcin dialect reflected
in the Cyrillic versions of 1902 Azbuka retained the liquid elision of Middle Korean.
(7) When the infinitive ending -e is attached to a stem ending in i, the form is usually shortened by
one syllable: ~ie -*■ -ye. The shortened form is more or less the written standard for polysyllabic
stems (kitalye *- kitalie ‘waits for’) but monosyllabic stems are usually spelled out in full (kie
‘crawls’) except for the unshortened copula (ye = ie ‘it is’) and the auxiliary verb ci- ‘become’ (-e eye
= -e cie ‘gets to be’ as in hulye eye ‘gets to be cloudy’). Except in special circumstances the phoneme
lyl does not occur after s, c, cc, or ch, but the recommendation of the orthographers is to regularize
the abbreviation -ie -* -ye in these cases, too, and write kasye (rather than kase) for kasie ‘someone
esteemed goes’, kacye (rather than kace) for kacie ‘holds, gets’, kaluchye (rather than kaluche) for
kaluchie ‘teaches’, sal-ccye (rather than sal-cce) for sal-ccie ‘gets fat’. What is said for the infinitive
ending -e also applies to the past-tense forms in (-i- +) -ess-: kasyess.ta (rather than kasess.ta) for
kasiess.ta ‘someone esteemed went’, kacyess.ta (rather than kacess.ta) for kaciess.ta ‘has got’,
kaluchyess.ta (rather than kaluchess.ta) for kaluchiess.ta ‘taught’, sal-ccyes.ta (rather than sal-
94 PART I
A Reference Grammar of Korean
ccess.ta) for sal-cciess.ta ‘got fat’. The endings -sye and -syess- are particularly frequent because they
contain the honorific marker -(u)si-, which can be used to turn almost any verb stem into an honorific.
Some speakers pronounced -sye not as /se/ but as /sye/ with the palatal sibilant they have learned to
make for foreign loanwords such as syassu/syaccu ‘shirt’, but I have not heard speakers make a
distinction of /eye/ from /ce/. The shortening of -ie to -ye is the source of virtually all cases of ty,
thy, and /tty/: titye = titie ‘treads’, pethye = pethie ‘props’, eph-tye /epttye/ = eph-tie (= eph-
tulie) ‘overturns’. Words which, in the spellings of earlier days, once had ty- are now pronounced c-
in the south and in the standard language, t- in the northwest dialects, such as that of Phyengyang,
which is famous for the word tengke-tang = cengke-cang ‘[railroad] station’ (from earlier tyengke-
tyang) - see §2.10.3. I have tried to follow the recommendation of the Korean Language Society in
this book, though I would prefer to have all the forms spelled out in full as -ie, both in the
Romanization and in Hankul, so as to avoid possible confusion.
(8) The dropping of h or ng (§2.7) sometimes leads to further shortening of vowel strings: ohilye -*■
(oilye -*■ ) oylye ‘rather, contrary to expectation’; kongyen hi -*■ ko’yen ’i -*■ koyni, kwayni ‘in
vain’; aymay han -*■ aymay ’(e)n -*■ aymen ‘vague’; siwen hata -*■ syen hata ‘is refreshing’.
(9) —’ tus hata = -nun tus hata (in literary cliches) ‘seems to do’; ’ mds hanta = -ci mos hanta
‘cannot do’ (in a few expressions).
(10) Final -i, at times itself a morpheme, drops from the first member of a number of compounds,
especially those involving diminutive suffixes: kkoli + -ayngi -*■ kkolayngi ‘tail’, kaci + -angi -*■
kacangi ‘branch’, kkochi + -ayngi -*■ -ayngi -*■ kkochayngi ‘skewer, spit’, thokki + -ayngi -*■
thokkayngi ‘rabbit’, ppuli + -eyngi -»■ ppuleyngi ‘root’; taykaii + -ppaii -» taykal-ppali, +
-ppayki -*■ taykal-ppayki ‘head’.
(11) A phrase with two similar syllables juxtaposed sometimes reduces the first: ec’ cenyek = eceyq
cenyek ‘last night’.
(12) The auxiliary adjective siph.ta appears in a shortened form taken from the southern dialect variant
siphuta. Although that variant itself is seldom heard in Seoul, the shortening found in -ko ’phuta ( =
-ko siph.ta) ‘wants to’ is quite common: nay ka poko ’phun sinmun ‘the newspaper I want to see’.
(13) For still other cases of shortening, see the various stems that are called s-dropping (§8.1.5),
ambivalent (or h-dropping, §8.2.4), and - from the viewpoint of the Hankul spelling - the
1-extending vowel stems (§8.2.2).
3.6. Vocabulary.
By source the bulk of Korean vocabulary falls into three classes, which I will call CORE or (even
though it may contain early loanwords) native-Korean, Chinese (systematically borrowed from China),
and English, though the class contains modern loanwords from other European languages. Many of
the modern loanwords were borrowed through their Japanese renderings, but some of those have been
given new versions taken directly from English. There remains considerable controversy over the
standardization of current loanwords from English. The trend is to favor close imitation of American
pronunciation of the words, rather than to follow Japanese patterns, as was sometimes done in the past.
But for words well established over several generations, the now traditional version is usually
conceded.
The Chinese vocabulary, which can be referred to as “Sino-Korean” when reference to it might be
confused with the language spoken in China, has been well integrated during the past thousand years
and it is now a component of the language - in sheer quantity the major component. It is interesting to
observe that while the majority of all words in a Korean dictionary are of Chinese origin, only about
ten percent of the words in the so-called “basic vocabulary” come from Chinese. Virtually all non-
Korean words have been brought into the language as uninflected words, as some kind of noun. When
the word clearly carries a verbal meaning, Korean has treated it as a verbal noun, putting the loanword
A Reference Grammar of Korean
PART I 95
into construction with a postnominal verb, typically ha- ‘do/be’. But there are also verbal nouns in the
core vocabulary, most conspicuously the mimetic words described in §14. I know of only two verb
stems of possible Chinese origin: sangwu- ‘harm’ (= sang ha-) ?< syang hwo- and pay- < poy-
‘conceive (a child)’, if that is from the Chinese morpheme pay < PHOY ‘fetus’ rather than pay < 'poy
‘belly’. (The non-Chinese etymology is supported by the Japanese derivation of harajnu < *para-ma-
‘get pregnant’ from hara n < *para ‘belly’; there is no alternative possibility from Chinese.)
Owing to the severe restrictions on syllable types in Chinese, morphemes of the Chinese
vocabulary are limited to a rather neat pattern of shapes, roughly those permitted by the chart of
Korean syllables spoken in isolation, with the exception of most of those with geminate initials (pp tt
cc ss kk). A list of all the occurring types of Chinese vocabulary will be found in §4.5; those shapes
that end in ™p -1 -k had final unreleased -p -t -k in Middle Chinese, as they still have in
Cantonese. The core vocabulary, on the other hand, includes some shapes which are less than a
syllable, such as the -n of chan ... ‘... that is cold’ from cha-, the -ss- of kass.e ‘went’ (from ka-) or
even less than a phoneme, as in the alternant of the infinitive that is represented by the palatal feature
(front component) reflected by our spelling -y in hay ‘does’ from ha-. The core vocabulary includes
some shapes which are more than a syllable but less than two syllables: kiph- in kiph.e ‘is deep’, pakk
in pakk ey ‘outside’, kkoch in kkoch ita ‘it’s a flower’, kaps in kaps i ‘the price [as subject]’. And it
includes some shapes which are two or three syllables (phulu- in phuluta ‘is blue’, meli ‘head’,
cwumeni ‘pouch’, kitali- in kitalinta ‘waits’) or something slightly more than two: muluph in
muluph i ‘the knee [as subject]’. Words of four syllables or more are usually either borrowings or
compounds (as are many words of two or three syllables), but in some cases the origins are obscure.
The alternations in shape of the Chinese vocabulary are fairly easily stated, as are the basic
shapes. For the Korean vocabulary the statements are more complicated because: (1) “overstuffed”
morphemes (like pakk and kiph-) must be reduced to permissible syllables when not followed by the
copula i- or a particle that begins with a vowel; (2) decisions must be made on morpheme boundaries
within words, and the decisions are not always so easy as they are for the Chinese vocabulary, where
we are helped by the restricted shapes and the morphemic writing system of Chinese; (3) there are
several special rules when endings are attached to verb stems.
Two morphemes, core and Chinese respectively, sometimes have shapes that begin with ng. These
are -ngaci ‘offspring of’ (ultimately perhaps from aki ‘child’, though we find no other cases of
affricating palatalization of the velar in a noninitial syllable) in songaci ‘calf’ (so ‘cow’), mangaci
‘colt’ (mal ‘horse’), and kangaci ‘pup’ (kay ‘dog’); and -nge ‘fish’ in fish names taken from the
Chinese: ocinge, punge, 'nonge, *inge, sange, swunge, kwange, and paynge < payk-(ng)e.
3.7. Layers of vocabulary in earlier Korean.
By the time Hankul was created Korean had acquired many borrowed words from various other
languages, such as Manchu and Mongol. Most of the loans, however, came from classical Chinese,
which was the standard written means of communication. The Chinese words were borrowed as
logographic characters and pronounced with an approximation to the Middle Chinese sounds. But some
of the words were borrowed early and got thoroughly nativized, so that their association with the
characters, and the traditional Chinese phonetic values, was forgotten. Most loans, however, retained
their association with Chinese even when they became part of the common vocabulary of speech, as
when san ‘mountain’ and kang ‘river’ displaced the native words attested as MK "mwoyh and ko'lom.
Scholars set up a system of somewhat artificial readings for the characters and codified this in a
dictionary of character readings that was published in 1448 under the title Tongkwuk cengwun
( ' twong-'KWUYK "CYENG-'ngwun - the tone mark on the first syllable is unexplained). The
Tongkwuk readings were an attempt to capture in terms of Hankul symbols the traditional phonetic
distinctions of Middle Chinese as found in the rime lists written by Chinese philologists. This
reconstruction of Chinese phonology took place nearly five centuries before the Sinologist Bernard
Karlgren interpreted the distinctions of Middle Chinese in terms of the Swedish Dialect Alphabet. The
Tongkwuk readings were written as Hankul syllables immediately following the corresponding Chinese
96 PART I
A Reference Grammar of Korean
characters in many of the texts of the 15th century. The modern way of pronouncing Chinese
characters used in Korea is simpler than the Tongkwuk readings in that certain unrealistic distinctions
(such as the initial velar nasal and glottal stop) are abandoned, and the unaspirated voiceless initials of
Middle Chinese are equated with the plain series of Korean obstruents, rather than the emerging tense
(reinforced) series favored by the prescriptive orthographers of the 15th century.
For certain words the prescriptive readings coexisted with nativized versions, so that a number of
doublet forms can be found in the texts. The nativized version often appears in Hankul without the
accompanying Chinese characters (Cf LCT 1971:78):
cin'sil Iwo (1481 Twusi 20:19b) = cin-'ssilq two (1482 Kum-sam 2:16a) = CIN-'SSHQ Iwo (1459
Wei 9:36d) ‘truly’
cywongywo (1518 Sohak-cho 8:9a cywong'ywo y *- cywongywo i; 1586 Sohak 3:8b) ‘essence, the
essential’ = cwong-'qywow (1482 Kum-sam 2:69a)
cywungsoyng (1447 Sek 6:19b, ...) = ' cywung-soyng (1447 Sek 6:5b, ... ) ‘creatures’
kanan (1475 Nay 2:2:47b; 1482 Nam 1:30b), kannan (1475 Nay 1:30a) = kan-nan (1447 Sek
6:15b, 13:56b; 1465 Wen 3:3:1:62a) ‘poverty; poor’
kuypyel (1445 'Yong 35, 1459 Wei 2:43a) ‘tidings, news, a letter’ = KUY-'PYELQ (1447 Sek 24:16a)
kwong'so (1447 Sek 9:30b) = kwong-'sso (1459 Wei 9:50b) ‘engaging in public affairs’
mi'hwok (1447 Sek 9:36b) = myey-'hhwoyk (1459 Wei 9:17b) ‘bewilderment’
"naycywong (1462 'Nung 1:20a, 1463 Pep 3:161a), "nay'cywong (1518 Sohak-cho 8:19b) = ~nay-
cywvng (1447 Sek 13:29b) ‘finally’
si'cyel (1475 Nay 2:2:47b, 3:32a; 1481 Twusi 7:25b) > si'cel (1518 Sohak-cho 8:19b, 21b) = ssi-
■CYELQ (1447 Sek 9:2a, 13:47b, 13:60b; 1462 ‘Nung 2:114b, 5:85b; 1459 Wei 18:83a; 1465 Wen
2:3:1:52a; ... ) ‘time (when)’
syang'nyey (1447 Sek 6:10a; 1459 Wei 10:7b, 17:35a; 1463 Pep 5:212b, 1464 Kumkang 64b, 1482
Kum-sam 2:25a, ? 1517- Pak 1:14b) = ssyang-'lyey (1447 Sek 9:14a) ‘always’
twocok (1445 'Yong 30; > twocek) = 'TTWO- CCUK (1459 Wei 2:19b) ‘thief’ (= totwuk)
tyangsyang (1466 Kup 2:64a) = ttyang-ssyang (1459 Wei 8:8b, [1447-*] 1562 Sek 3:22b) ‘always’
"wen °ho- ( ? 1517- Pak 1:60a) = "ngwen °ho- (1586 Sohak 6:44a) = ‘ ngwen °ho- (1447 Sek
9:40b, 13:44b, 24:18a; 1459 Wei 13:35b)
wuytwu (1447 Sek 13:6a) = ngwuy-TTWUW (1459 Wei 10:25a) ‘forming the head/van’
"wuy °ho- (1447 Sek 6:1a, 6:7b, 6:13b, 6:16a, 13:49b) = 'wuy °ho- (1459 Wei 7:17b, 9:5-6,
13:35b, 13:36a, 17:54a; 1463 Pep 2:172ab, 2:231b, 7:17a; 1465 Wen se:6a; 1482 Kum-sam 5:48-9)
‘do for (the sake of)’
LCT 1971:78 finds over thirty words that were usually written without the appropriate Chinese
characters, and presumably they were all well assimilated. Additional notes on some of those words:
"camskan (1459 Wei 7:15b) = "cams.kan (1447 Sek 13:53b; 1475 Nay 1:55b, 2:1:2b; 1463 Pep
2:226a; 1465 Wen 1:1:1:44b; 1481 Twusi 7:1b; 1482 Kum-sam 2:13b; 1482 Nam 2:31a) ‘a while’ <
(1*)'ccam s kan “the space of a while”
'cukcay (1459 Wei 9:35f; 1463 Pep 6:97a; 1466 Kup 2:4b) < 'cukca 'hi (1447 Sek 6:2a,
6:11a, 9:12b, 24:16a; 1459 Wei 2:6b) ‘suddenly’ < *'cukca [< *'cho ] hi
coycwo (1447 Sek 6:7a) ‘talent, ability’ < *CCOY-CHWOw
"cwosim (1447 Sek 9:37a, 1459 Wei 1:6a) ‘taking care’ < *chwow-sim = Beijing caoxin
cyang'cho (1459 Wei 1:18a, 17:78b; 1462 'Nung 1:28b, 7:73b; 1463 Pep 1:123a, 3:35a; 1475 Nay
2:1:30a; 1482 Kum-sam 4:22b) ‘in future’ < *cyang-'CHO (> Beijing jiangci ‘for a while; almost’)
"hoyng'tyek (1447 Sek 6:2b) ‘deeds’ < (?*) hoyng ' tuk “perform virtue” (Cf Soothill 221b;
LCT and Kim Wancin assign the second syllable to two different characters but both are read ' cyek )
in 'so (1459 Wei 2:9a) ‘greeting (bow)’ < *ZIN-'so “people-thing” (> Beijing renshi ‘gift’)
kwu'kyeng (1459 Wei 2:27b, 2:35b, 7:11a) ‘viewing’ < *'kwuw-"kyeng “seek the scene”
kwuy-s kes (1447 Sek 6:19b, 1482 Kum-sam 2:7b, 3:27b, 3:34b) < "KWUY s kes “devil(’s) thing”
A Reference Grammar of Korean
PART I 97
"Iwongtam (1447 Sek 6:24b, 1459 Wei 1:44b) ‘joke’ < *'LWONG-TTAM “play talk”
"moyzyang (1463 Pep 2:189b; 1481 Twusi 7:2b, 22:1b) > "moyyang (1481 Twusi 15:20a, 1518
Sohak-cho 10:12b) ?> 'moy'yang (id. 9:24b), moyyang (id. 10:1b) ‘always’ < "Moy (s) 'yang
“each(’s) appearance”
phwunglywu (1447 Sek 9:21a, 13:9a) ‘music’ < *pwong-LYWU (Cf Beijing fengliu ‘elegance’)
pwun'pyel (1447 Sek, 9:29b, 13:36a; 1459 Wei 2:6a) ‘thinking, considering, worry’ < *pwun-
'(P)PYELQ (> Beijing fenbye ‘separate, distinguish’)
si'hwok (1462'Nung 9:88a, 1463Pep 6:145b, 1481Twusi 8:8a) ‘sometimes; perhaps’ < *ssi-'H\vok
syang toWoy- (1459 Wei 1:43a) ‘is common, vulgar’ < ssyang ‘constant, always’
tangta.ng-i (1447 Sek 19:34a, 1459 Wei 1:7b, 1466 Kup 2:64a, 1481 Twusi 7:4b, 1482 Kum-sam
4:20b) = tangtang-i (1481 Twusi 7:9a, 7:31a, 20:34b) < *tang-tang hi ‘suitably’
"tyangka (1447 Sek 6:16b, 6:22a) ‘[marriage into] the husband’s family’ < *"ttyang-ka
For cyen hye, "hoyng hye, and 'twok 'hye see Part II, hye. LCT’s list is representative, but not
exhaustive. We can add, among others:
namcin (1459 Wei 1:43b) ‘male; husband’ < *nam~zin “male person”
nam 'phyen ‘husband’ < *Nam~ppyen “male side /direction /party”
"sya wong ‘husband’ < * sya-wong “house elder”
kansywu °ho- (1447 Sek 9:36a, 1465 Wen 2:3:2:88b), kanso °ho- (1475 Nay 3:32a) < *(')KHAN-
"SYWUW ‘guarding’ > Beijing kanshou
'tyang'so ‘selling’ (1459 Wei 13:8b, 23:64a; 1463 Pep 6:170b) < *ttyang-'so “market event”
Several common elements of modern Korean have Chinese origins that are now largely forgotten:
- ca ‘person’ < "CYA, ■■■ cha ‘on the verge; (as) an incidental consequence of; (for) the purpose of
< 'CHO ‘next, second(ary)’, ... .
Certain characters were given more than one reading (§4.7), sometimes reflecting divergent
meanings in classical Chinese, and this led to doublet compounds, as well: "KAY-'THWALQ (1459 Wei
14:39b, 1462'Nung7:27a, 1482 Kum-sam2:4ab) = "HHAY-'THWALQ (1447 Sek 13:43b) ‘emancipation’.
Here the character itself represents a triplet, with a third reading 'HHAY as in 'MYWOW-'hhay (1482
Kum-sam 5:24b) ‘the wondrous understanding’. Doublet forms that differ only by accent are virtually
unfound in the Chinese part of the vocabulary. Tongkwuk cengwun recognizes more than one tone for
certain characters: ‘dye’ is listed as zyem, "zyem, or 'zyem, but the only example I have found is the
last, in the expression 'zyem-'CCYeng (1465 Wen 2:3:1:43a) ‘pure-washed’. The character for
‘separate, special’ was read both pyelq and ppyelq : 'kak-'pyelq hi (1447Sek 13:10b) = 'kak-
PPYELQ hi (1462'Nung 1:89a, 1482 Nam 2:63b) ‘especially’.
Indirect evidence that the usual reading of mwon ‘gate’ was mwun (as made explicit in 1527
Cahoy 2:4a = 7a) can be found in the choice of variant forms of particles in these passages: spol'li
mwon ul "yel'la ho 'ye ’ys.te'n i (1459 Wei 10:25a) ‘wanted to open the door quickly’ (we expect
mwon 'ol ); kasoy 'Iwo 'hwon [mwon] ulan sywokcyel ["Jep.si [~]ye[']ti ["]mal[']la (1481 Twusi 7:9a)
‘do not to your regret open a door made of thorns’ (we expect mwon 'olan ). Incorporated in the word
SA-MWON ‘sramana’: sa-MWON 'uy "swon-'toy (1447 Sek 24:22a) ‘to the sramana’ (we expect 'oy). In
sep [MWON] 'ey (1481 Twusi 7:9b) ‘to the twig gate’ the writer was probably following the nativized
mwun that would have been the normal spoken version, as attested in Hankul somewhat later ( ? 1517-
Pak 1:12b), for otherwise we would expect the particle to be ay.
Some of the compounds may have been borrowed from Chinese dialects of the day:
'cokya (1445 'Yong 25, 1447 Sek 6:5b, 9:33a; 1459 Wei 13:8b) < *'cco-ka = Beijing zijia ‘self’
phunco ( ? 1517- 'No 2:23, 26) < *"pwun-"CO = Beijing fenzi ‘starch’
sywu'lwup (1446 Hwun 29a [the dot is missing in the Taycey-kak repro but clear in the photocopy
of 'Yi Sangpayk 1957]) ‘umbrella’ ?< *"SYWUY-'LIP = (?) dialect equivalents of Beijing shul ‘water’
+ li (< *lyep) ‘umbrella’
yang'co (1447 Sek se:5a, 6:13b, 23:4a; 1459 Wei se:16a, 8:15b, 8:19b, 23:86b; 1462 'Nung 3:84a)
98 PART I
A Reference Grammar of Korean
< *' yang-" co = Beijing yangzi ‘appearance’; Cf the postmodifier "yang < ’yang ‘pretense’.
But other compounds were possibly made up in Korea and then nativized. That appears to be the
explanation for cyen'cho (1447 Sek 6:2b; 1459 Wei 7:13b, 9:35d; 1462 'Nung 1:64b, 1:77b; 1465
Wen 1:1:1:47a; 1463 Pep 5:169b; 1482 Nam 1:5a) ‘cause, reason’ < cyen ‘effect’ + cho ‘next,
second(ary)’; perhaps "chyen-lyang (1447 Sek 9:13a, 24:47b; 1463 Pep 6:144a; ... ) ‘money and food’
< CCYEN ‘money’ + LYANG ‘provisions’ (modern Beijing qianliang means ‘tax; husband’s allowance’)
and (canchi <) 'can'choy ‘banquet’ < chan ‘meal/eat’ + "choy ‘vegetable’. The noun hwe (1462
'Nung 6:96b, ... ) ‘boots’ is taken from an ancestor of Beijing xue (‘boots’, not xie ‘shoes’) that is more
immediate than the form reflected in the reading HWA.
Some of the Buddhist terms are borrowed from Chinese transcriptions of Indie words: MI-’LUK
‘maitreya (the Buddha to come)’, ’NYELQ-PPAN ‘nirvana (extinction)’, ’ppikhwuw ‘bhiksu, (almsman,
mendicant monk)’, ' ppi-khwuw-ni ‘bhiksuni (nun)’, sa-mi ‘sramanera (religious novice)’, sam-ma-ti
(1462 'Nung 5:31b) ‘samadhi [a trance-like state of unperturbable meditation]’, sam-’moy ‘samadhi
(meditation)’, sa-Mwon ‘sramana (begging monk, ascetic)’. Others are Chinese caiques (loan
translations) of Indie expressions: ' cyeng-’KAK ‘sambodhi (Buddha wisdom)’, ' cywung-soyng ‘sattva
(all living things)’, ' kyelq-’cip = ‘samglta (a council to consolidate and collect the Buddha’s teachings
and to decide orthodoxy)’, ’PEP ‘dharma, (Buddha’s) law’, syeng-MWUN ‘the sravaka (= hlnayana
disciple in the first stage)’, "sywow-ssing ‘hlnayana (the Lesser Vehicle)’, ’ttay-ssing ‘mahayana (the
Greater Vehicle)’, "ttwow ‘marga (the Way)’, ywen ‘pratyaya (secondary cause)’.
A number of words sound as though they might be from Chinese, but no characters have been
associated with them, e.g. yengmun ‘reason’ (no early attestations?), "ywumwu (1447 Sek 6:2b, 1449
Kok 61) = "ywu'mwu (1518 Sohak-cho 8:22a) ‘a letter’, "cywong ‘slave’ (?< cywong ‘follow,
obey’), ... . Certain words often suggested as Chinese loans may have other origins. Kang Hengkyu
1988:192 takes soyng'kak ‘thought’ to be from Mongol sanaga rather than Chinese "soyng
‘contemplate, recollect’ + ’KAK ‘awaken’ (LCT 1971:78, with question mark), and solang ‘thought >
love’ to be connected with solh ‘flesh’ (and/or "sal- ‘live’? - the vowel is disconcerting) rather than
from so-LYANG (LCT 1971:87) = Beijing slliang ‘consider’, but those etymologies seem less
convincing than the Chinese compounds. The noun nungkum ‘apple’ (first attested ? 1834-) is from
lingkum ( ? 1517- Pak 1:4b) < LIM-kkum. Both "cywung ‘monk’ and 'swung ‘nun’ seem to be variant
forms of sung ‘Buddhist priest’. The expression 'sywok'cyel "ep'si ‘in vain, futilely’ seems to be from
SYWOK-’CYELQ "ep's-i ‘without (even) a brief religious ceremony’.
Kim Wancin (1971:228-30) lists 36 words that he considers to be prehistoric borrowings from
Chinese, which preceded the wholesale borrowings called Sino-Korean, referring to reconstructions of
Middle and Old Chinese readings of certain characters. Some of the words are well deserving of such
attribution: 'cah ‘measure’, sywoh ‘vulgar, lay(man)’, tyeh ‘flute’, and zywoh ‘mattress’ are surely
Chinese loanwords, as is mek ‘ink stick’. The nouns pwut ‘writing brush’, pwuthye ‘Buddha’, and
tyel ‘temple’ are to be considered together with their Japanese counterparts as cognate borrowings. The
derivation of cek ‘time’ from Old Chinese *dyeg (> Beijing shi) is intriguing but must be weighed
against the comparison with Japanese tokf, just as the derivation of "kwom < kwo'ma ‘bear’ from Old
Chinese *gyum must be weighed against Japanese kuma 1 , and the derivation of talk ‘chicken’ from Old
Chinese *tyeg ‘bird’ must share attention with the putative Japanese cognate tori" 1 . Other derivations
that look good include 'sal ‘arrow’ from Old Chinese *Syer and 'pwoy ‘hemp cloth’ from Middle
Chinese *pwo' ‘cloth’ (Kim Wancin mistakenly labeled this shape as Old Chinese, but that form was
*pwag). On the other hand, mol ‘horse’ is more likely to have been directly borrowed from Mongol
mori rather than an early Chinese equivalent of "MA, as Kim Wancin proposes.
4.0. Shapes.
Morphemes are abstract entities that take on shape only when they are realized as what are
sometimes called morphs, just as phonemes take on substance only when they are articulated as
PHONES. When a single phoneme is articulated with perceptibly different sounds, often determined by
the environment, the phones are said to be ALLOPHONES of that phoneme. When a morpheme is
A Reference Grammar of Korean
PART I 99
realized in more than one shape, usually depending upon the adjacent sounds or morphemes, the
morphs are said to be allomorphs of that morpheme. Quite often the variation in the shapes can be
described in general terms that apply to groups of morphemes or to types of shape. Words and stems
often contain more than one morpheme, and they too can also be described in terms of shape types.
4.1. Shape types.
A word, or a morpheme, sometimes occurs in more than one shape. The shape is the way the
element is represented in phonemes, as actually pronounced. In general we find a resemblance between
the several shapes of a given element. Except for the nominative particle i/ka and a few of the
inflectional endings, the different shapes of a given morpheme or word have some stretches of phonetic
makeup in common. The differences in shape between the alternants of many morphemes and words
can be stated in general terms; such statements are often called morphophonemic rules. Some of
these rules were stated in §2.6; a glance at the chart of permitted consonant clusters tells us to expect
that a morpheme which sometimes has the shape sip can be expected to show the shape sim (and even
the shape si) in certain environments. These alternations are so automatic (every “expected” -p before
a nasal turns out to be pronounced -m) that for the most part they are disregarded by the Hankul
spelling and by the Yale Romanization. When we hear the spoken sequence •mm- we cannot be sure
that it will be spelled p + m (or ph + m, ps + m) instead of m + m unless we recognize the
morphemes or words involved. As a result of convergence, Korean has a fair number of words and
phrases that sound the same but are spelled differently because each constituent part is always written
according to its “basic” shape, as found in some of its other environments. That is why the string
/cimman/ is written cim man when it means ‘just the burden’ but cip man when it means ‘just the
house’ and ciph man when it means ‘just the straw’.
The alternations shown in the cluster table are automatic in that you apply them to shapes
regardless of the particular words involved; and they are fully automatic because you need not
even know what the grammar of the words may be. A similar kind of fully automatic alternation is
found in English when an expected “s” is pronounced /z/ after Igl: we do not have to know the
grammar to pronounced “legs” as /legz/, for the rules of our language automatically keep us from
saying /legs/. Such rules work for nonsense words as well as real words: the pseudo-word “blegs” can
only be pronounced /blegz/.
Notice that the morphophonemic rules apply only if the two syllables are run together, with no
pause intervening. In general, that is true for most of our rules for alternations that occur at the point
of contact between Korean morphemes or words. Thus, 1 + n -* /11/ in ku tul ney /kutulley/ ‘they’,
tal nala /tallala/ ‘the moon (as a place)’, saynghwal-nan /saynghwallan/ ‘the hardships of life’, tul-
nol.i /tulloli/ ‘picnic’, and other expressions where pause virtually never intervenes. In certain other
expressions, pause is infrequent: kaul nal /kaullal/ ‘autumn day/weather’, onul nal /onullal/ ‘today’
- compare onul nal(-ssi) /onul I nal(ssi)/ ‘today the weather » ’. The expression cal nol.a ‘plays
nicely’ (adverb + verb) is usually pronounced without pause /callola/. Pause is infrequent between
short unmarked object and verb, so that atul nah.ko ttal nah.ko ‘giving birth to sons and daughters =
(lived) happily ever after’ is usually said /atullakholttallakho/. With a marked object, there are
common versions with or without pause: al ul naynta ‘lays eggs’ may be heard as /alullaynta/ or as
/alul I naynta/. Since the accusative particle ends in 1 and the verb nay- ‘puts out, ... ’ takes many
different nouns as object, that verb is frequently heard in the alternant shape lay-. There are al§o cases
of unmarked subject + verb that are such common expressions they are usually said without an
intervening pause, e.g. pul nanta /pullanta/ ‘a fire breaks out’ (= pul i nanta). Kyel nanta and kol
nanta, both meaning ‘gets angry’ (from ‘temper appears’), are usually pronounced /—11—/ as are their
synonyms kyel naynta and kol naynta (‘displays temper’). If a pause is inserted, it would be more
natural to attach the appropriate particle to mark the subject or object: kyel i I nanta, kyel ul I
naynta. In expressions of modifier + nominal, an intervening pause is usually unnatural in relaxed
speech, so that tte-nal nal ‘the day to leave’ is usually pronounced /ttenallal/. We can know the
100 PART I
A Reference Grammar of Korean
appropriate spelling for the expressions -ul nalum ita ‘it depends on’ and -ul nawi (ka) eps.ta ‘there is
not enough to; there is hardly a need to’ only from etymology or reading pronunciations, for they are
usually pronounced with /—11--/. To be sure, by inserting a somewhat artificial pause the Korean
speaker can distinguish an otherwise homophonous phrase like salq kos ‘places to buy (them)’ from sal
kkoch ‘flowers to buy’, both /salkkot/ in normal speech, and mos kanta ‘can’t go’ from mos kkanta
‘can’t peel’, both /mokkanta/ in normal speech.
The only cases of fully automatic (phonemically determined) alternation other than those from the
table of permitted clusters are (in part) the alternations of 1 and n (§4.4) when after pause, and of yey
(§4.3) when not after pause. Other alternations are widely but not fully automatic, because you have to
be aware of at least some grammar to decide whether they apply. In the following sections six kinds of
alternations are described:
(1) Treatment of syllable excess, §4.2.
(2) Treatment of yey, §4.3.
(3) Treatment of 1 and n, §4.4.
(4) Occurrence of reinforcement (-q) with the prospective modifier -ul, §1.5, §9.3.
(5) Occurrence of reinforcement (-q) with consonant stems that end in sonants (m, n, or an 1 that is
reduced from basic Ip, lph, 1m, 1th, lk), §8.1.1.
(6) Various alternations of “two-shape” elements, §5.1.
4.2. Syllable excess.
There is a limited group of “overstuffed” morphemes, each of which has a basic form that ends in
a consonant that can occur only at the beginning of a Korean syllable or in a string of consonants that
can occur only if divided between two syllables, part at the end of one syllable and part beginning the
next. The “overstuffing” or syllable excess is heard only before certain vowels. Before pause or a
consonant - and in certain constructions also before a vowel - the excess is replaced by those
corresponding consonants which are permitted at the end of a syllable. Before certain consonants the
excess replacement then undergoes further replacements, those that are phonemically determined for
the consonant (§2.6). For example, kaps ‘price’ is reduced to the shape kap before pause or in phrases
like kap to ‘the price too’; the final p of this shape kap is then subject to the automatic alternations of
any final p, so we hear /kamman/ for kaps man ‘just the price’.
Before a vowel which (1) begins a particle, such as the nominative marker i or the accusative ul,
(2) begins the copula i-, or (3) begins an inflectional ending, such as the infinitive -e or the adversative
-una, the full basic form is heard with its syllable excess intact: kaps ul ‘the price [as object]’, kaps i
‘the price [as subject]’, kaps ita ‘it’s the price’. (The phonemic shapes are actually /kapssul/ and
/kapssi(•••)/, because of the automatic rule under which an orthographic ps is not distinguished in
pronunciation from pss, as we earlier observed.)
Before a vowel which does not begin an inflectional ending, the copula, or a particle, the usual
treatment reduces the excess: kaps olumyen ‘when the price rises’ is pronounced /kapolumyen/ and
kaps alki elyewe ‘it’s hard to find out the price’ is pronounced /kapalkielyewe/. There are exceptions
in a few compounds (yetelp hay ‘eight years’ is /yetelphay/), in derived verb forms (olk.hi- ‘get
roped’ /olkhi-/), and in iterated noun + the adverb-deriving suffix -i (moks.moks-i ‘in portions, in
shares’ is /mongmokssi/). In a few combinations both treatments occur: /masisse/ or /matisse/ for
mas iss.e ‘is tasty’ but the former is better regarded as mas i ’ss.e, a reduction of mas i iss.e, as we
had occasion to remark earlier. According to one study (Kim Hyenglyong 1985) in modern written
Korean there are 1,757 different orthographic syllables that carry a “final” component (pat.chim), and
1,384 (= .787) of these carry codas that are allowed at the end of a phonetic syllable: p t k m n ng
1. The remaining 373 (= .213) represent morphemes with syllable excess.
The following list of morph-final strings includes all the occurring types of syllable excess. Some
of the types occur with both nouns and verb stems; others only with one or the other. There are also
A Reference Grammar of Korean
PART I 101
stems ending in h, lh, nh, and w, for which see §§8.2.2-3. Historically, there are nouns that once
ended in h, lh, nh, and mh, but they have dropped the final h in modern Korean. The etyma have left
morphophonemic relics in the case of salh ‘flesh’, anh ‘inside’, amh ‘female’, and swuh ‘male’, but
the words in which a reflex of the h appears are now spelled (with respect to this feature)
phonemically: salkhoki = salh-koki ‘red meat’, anphakk = anh-pakk ‘inside and outside’, swukhay
= swuh kay ‘male dog’, amkhay = amh kay ‘female dog’. (See below for more on this. A list of the
MK -h nouns is at the end of this section.) There are also relics of excess at the beginning of certain
syllables: pssi = ssi ‘seed’ in pyepssi = pye-pssi ‘rice seed’, pssal = ssal ‘grain’ in ipssal = i-pssal
‘raw rice’, pttu- = ttu- ‘open (one’s eyes/ears)’ in chiptte = chi-ptte ‘looking with raised eyes’, ... .
For the inflectional stems ending in -1-, which show different behavior from other elements ending in
1, see §8.1.4. There are a few archaic examples of excess mk in nouns: namk = namu ‘tree’ (Cf
modern namak-sin ‘wooden shoes’), kwumk = kwumeng ‘hole’. These go back to MK nouns that
had two allomorphs which developed from *-m u fok\ there were similar types from *-n %k, *-s%k,
*-/%&, and *-/%/. Those nouns are listed in §8.3.1, where we see how the verb stems of this type
developed into the peculiar alternation found in the modern -ll- verbs. For all of the nouns with more
than one shape, including those with syllable excess, the free shape that occurs before pause is also
used before certain particles, such as the MK genitive s.
The first column of the list shows the morph-final ending, the second column shows the phoneme
to which the excess is reduced; the third column gives a noun example, and the fourth a verb example.
The notes immediately follow the list.
LIST OF MORPH-FINAL STRINGS
p
cip ‘house’
cap- ‘catch’
tl
nat ‘grain’
tat- ‘close’
k
mok ‘throat’
mek- ‘eat’
1
mal ‘horse’
tul- ‘listen’ 2
m
kam ‘persimmon’
kam- ‘shampoo’
n
an ‘inside’
sin- ‘wear on feet’
ng
khong ‘soy bean’
-
th 3
t
path ‘field’
math- ‘take charge of’
s
t
os ‘clothes’
pes- ‘take off, doff’
ss
t
-
iss- ‘exist, stay’
C 4
t
nac ‘daytime’
chac- ‘look for’
ch 4
t
kkoch ‘flower’
coch- ‘follow’
ph
P
aph ‘front’
ciph- ‘lean on’
ps
P
kaps ‘price’
eps- ‘not exist’
kh
k
puekh ‘kitchen’
-
kk
k
pakk ‘outside’
kkakk- ‘cut, mow’
ks
k
moks ‘share’
-
Is
1
tols ‘cycle; postnatal year of age’
-
1th
1
-
halth- ‘lick, taste’
lk
k, (1)
talk ‘chicken’
ilk- ‘read’
lm
m, (1)
[salm ‘life’]
kwulm- ‘go without food’
IP
P, 0)
(yetelp ‘eight’)
palp- ‘tread on’
lph
P, (0
-
ulph- ‘intone, chant’
nc
n
—
anc- ‘sit down’
1 Many speakers treat the few nouns ending in a basic t as if they ended with an s. Even tikut ‘the
letter T’ is pronounced with final s by most speakers when it is followed by, say, the nominative
marker i. But the Hankul spelling writes final t for this and for a few other nouns. Choice of final s
102 PART I
A Reference Grammar of Korean
instead of t for the basic form of certain words, such as ches - ‘first’, would seem to be arbitrary,
or based on the notion “when in doubt treat final It/ as if from s”. The only basis for writing the
adverb mos ‘not possibly’ with a final s rather than the t used in older spellings is the word calmos
‘mistake’, derived from cal mos hanta ‘cannot do it well’ /cal(I )mothanta/, which is treated as
having a final s (calmos ul hanta ‘makes a mistake’).
2 But the consonant-stem tut-/tul- ‘hear’ is never pronounced with the syllable-final -1, in
contrast with the 1-extending vowel stem tu-1-, which has the syllable-final allophone before a
consonant, as in tulko ‘entering’ (compare tut.ko ‘hearing’). The infinitive tul.e is said as tu-le,
with the flap allophone, whether it means ‘hear’ or ‘enter’. See §8.3.2.
3 In Seoul th + i, or t + hi, is regularly replaced by /chi/: path ita is pronounced /pachita/ ‘it’s
a field’ though path ey is /pathey/ ‘in the field’, and the passive forms ket-hi-, tat-hi-, mut-hi-,
ppet-hi- are usually pronounced with /-chi-/. Notice that a morpheme boundary is always
involved; there are no cases of /thi/ within a morpheme. In a similar fashion, t + i is replaced by
Id/: /kwuci/ for kwut.i ‘firmly’ (but kwut.e ‘is firm’ is pronounced /kwute/); /ttampaci/ for ttam
pat.i ‘sweatshirt’ (but ttam ul pat.e ‘receives sweat’), /haytoci/ for hay tot.i ‘sunrise’ (but hay ka
tot.a ‘the sun rises’), /mitaci/ for ml-tat.i ‘sliding door’ (but mllko tat.e ‘pushes and closes’).
4 Many southern speakers treat noun-final c, ch, and th as if s: /nasey/ instead of /nacey/ for nac
ey ‘in the daytime’, /kkosi/ instead of /kkochi/ for kkoch i ‘the flower [as subject]’, /pasi/ instead
of /pachi/ for path i ‘the field [as subject]’ - and the pronunciation /pathi/ is heard in the north.
The reduction of the excess is as follows. A string of more than one consonant simplifies to one,
by dropping all consonants in excess of the first, with the exception of certain cases involving the
liquid and an obstruent. The strings Is and 1th act like most clusters, dropping all but the I. The string
lp also acts this way for the one noun example: yetelp reduces to /yetel/. Noun-final lk, however,
reduces to k, so that talk becomes /tak/. For verb stems the strings lk, 1m, lp, and lph are given both
treatments as competing variants. The standard variant seems to treat the liquid as excess, so that ilk-,
kwulm-, palp-, and ulph- are reduced to ik-, kwum-, pap-, and up- before adding an ending that
begins with a consonant, such as -ta or -sup.nita. But some people retain the liquid, so that the
reduction is to il-, kwul-, pal-, and ul-. Those who use the compound adjective yelp-pulk- ‘be light
red’ seem to pronounce it /yelpulk-/. Stem-final lk is most commonly treated in the standard way (with
the 1 dropping) except when attached to endings that begin with k, where the other treatment seems
more common: ilk.ko and ilk.ki are pronounced /ilkko/ and /ilkki/ rather than the /ikko/ and /ikki/
that (automatically compressed from ik-kko and ik-kki) we would expect as consistent with /ikcci/ *-
ilk-ci and /ingnun/ ( *- ik-nun) *- ilk-nun. The proper analysis of these forms is il- + /kko/ (etc.),
the reinforced (-q) allomorph of -ko, rather than ilk- + -ko, since endings regularly reinforce after a
liquid reduced from a cluster. Compare the unexpected treatment of salk ‘leopard cat’ + kwayngi ‘cat’
-» salk kwayngi /salkkwayngi/ ‘leopard cat’ where we expect /sakkwayngi/ as consistent with talk
koki /takkoki/ ‘chicken meat’. In overprecise speech, a theoretically dropped liquid sometimes
reappears, giving anomalous syllable-final clusters, as in /talktto/ for /taktto/ = talk to ‘chicken too’.
That is somewhat similar to the retention, or reimposition from spelling, of III by certain English
speakers in words such as “palm” and “calm”.
After dropping any excess, if what remains is not a permissible syllable-final consonant (p t k 1 m n
ng), as with s ss c ch, or if it is an 1 which is the last phoneme of a consonant stem (§8.1.4), but not
an 1 reduced from a cluster, that remaining consonant is treated as the phoneme Itl with whatever
reflex would be appropriate to t. (But historically the 1 of the -T/L- stems is a lenited form of t.)
Below is a fairly complete list of examples for each extrasyllabic final. But instead of s, for which
there are a large number of examples (as there are for p k m n ng 1), those examples ending in a basic
t are listed, since their number is much smaller. In each list, all the nouns are grouped at the end.
A Reference Grammar of Korean
PART I 103
EXAMPLES OF EXTRASYLLABIC EXCESS
t et- < "et- ‘obtain’, it- ‘be good’, ket- < ket- ‘fold up’, kot- < kwot- ‘be straight’,
kwut- < kwut- ‘be hard, firm’, mit- < mit- ‘trust, believe in’, mot- [obsolete] < mwot-
‘gather up’, mut- < mwut- ‘bury’, mut- < mwut- ‘stain, color’, nat-/nath- [obsolete] < nat-
Inath- = nathana- < na'tha °na- ‘appear’, pat- < pat- ‘receive’, pet- < pet- ‘stretch out
(like a road)’, ppet- ‘extend, stretch out (an arm or a leg)’, ssot- < 'swot- ‘pour out’, tat- <
tat- ‘close’, tit- (a truncation of titi- < tu'tuy-) ‘tread, step on’, tot- < twot- ‘sprout, bloom;
(sun/moon) rise’, ttut- < ptut- ‘bite, snatch, graze’;
kot < 'kwot ‘immediate, direct; to wit’, 'kwot > kwos ‘place’, mat < mot ‘senior,
eldest’, nat < "nat ‘grain’, ''pet > pes ‘friend’, 'tet > tes ‘a while’,
th cith- < tith- ‘be saturated, (liquid) thick, (color) dark’, 'heth- ‘get disarrayed/
scattered’, huth- < 'huth- ‘get dispersed, scatter out’, kath- < 'koth- < 'kot °ho- ‘be like; be
together’, kith- ‘remain’, math-j < math- ‘smell, sniff (it)’, math- 2 < mast- ‘take charge of’,
nath-/nat- [obsolete] < nath- / nat- = nathana- < na'tha °na- ‘appear’, path- < path- 1
‘sift, drain; (liquid) dry up’, payth- < path -2 ‘spit out’, puluth- (truncation of puluthu- <
pulu'thu-) ‘get swollen’, puth- < puth- ‘stick, be attached’, yath-/yeth- < yath-lyeth- <
nyeth- ‘be shallow’;
hoth < hwoth ‘single’, keth < kech ‘surface, shell’, khongphath < khwong'phoch
‘kidney’, koputh (= koputhangi) ‘outside fold of a bolt of cloth’, kkuth < (••• s) 'kuth ‘end’
(but kkuthu in kkuthu-meli ‘butt-end’), kyeth < kyeth ‘side’, melimath < (me 'li mathJ
‘head (of bed or grave)’, mith < mith ‘bottom, underside’, muth < mwuth ‘land, shore’,
nath < "nath ‘piece, unit’, pakkath < pas[k] kyeth ‘outside’, path < path ‘dry field,
garden’, phath < phoch ‘a kind of red bean’, pyeth < pyeth ‘(sun)light’, sath < sath
‘crotch’, soth < swoth ‘pot’, swuth < ( l*)swusk (CF Kim Thaykyun 323b) ‘quantity, bulk (as
of hair)’, toth [obsolete] < twoth ‘boar’.
ss iss- < is ft)- ‘exist, stay’, -(%)ss- (past) < is(i)-, -keyss- (future) ? < - 'ke ’ys(i)'ta.
There is also ssayss- ‘be plentiful’, contracted from ssah.ye iss- ‘be piled up’, which lacks the
expected modifier form (*ssass.un), replacing it with ssayn, contracted from ssah.in. The
modifier form of iss- is relatively uncommon but it occurs: see the entry iss.un in Part II.
c aykkwuc- ?< *"ay s kwuc- ‘be undeservedly misfortunate’, cac- ‘(wind) ease up, calm
down’, cac- < coc- ‘be frequent, incessant’, cac-/cec- ‘lean back’, cec- < cec- ‘get wet’, cic-
< cue- ‘(dog) bark, bay’, ccic- < peue- ‘tear it’, ccoc- ‘twist (a pigtail)’, cls-kwuc- ‘be
annoying’, chac- < choc- ‘look for, find’, ic- < nic- ‘forget’, ic- ‘wane; get chipped’, kac-
< 'koc- ‘be prepared’ (rare except in the causative kac.chwu- < ko'chwo- = koc-'hwo- ‘make
ready’), kac- (a truncation of kaci- < ka'ci-) ‘possess’, kkoc- < kwoc- ‘insert’, kkwucic- <
kwu'cic- / kwu'cit- ‘scold (a child)’, kwuc- ‘be bad, vile; (weather) be threatening’, (nwun i)
kwuc- ‘go blind’, mac- < mac- ‘be suitable, appropriate’, mac- < mac- ‘meet; face’, mayc-
< moyc- ‘bind, tie’, mec- [dialect?] < mec- = memchwu- < me'chwu- = memul(u)- (<
me'mul-) ‘stop’, mec- ‘be bad’, nac- < nac- ‘be low’, nuc- < nuc- ‘be late’, peluc- ‘scatter,
dig out’, pic- < pic- ‘brew, ferment, make’, putic- ‘bump into’, pulu-cic- ‘cry out, shout’,
seluc- < selec- ‘discard’ (obsolete), selkec- < selGec- (obsolete) = selkeci ha- ‘do dishes’,
tac- (truncation of taci- ?< [dialect] tati-) ‘harden by stamping, press, mince’;
cec < 'cyec ‘milk’, coc ‘penis’, i'su'lac ‘(wild) cherry’ > isulach = isulac(h)i
[dialect], kac’ (truncation of kaewuk < kach, used as adnoun) ‘leather’, kalac (truncation of
kalaci < kalati ? 1834-) ‘foxtail (plant)’ (= kangaci phul), kic ‘coat collar; portion’, kwoc\
(> kkoc.i) ‘skewer’, koc < kwoc 2 ‘cape, promontory’ (postnoun), koc’ (truncation of kocang
‘place’, Cf kos < kwot ‘place’), mec ‘cherry’, nac < 'nac ‘daytime’, nuc ‘late’ (adnoun),
nuc [obsolete] > nuch ‘sign, portent, omen’, on-kac’ < 'won-'kas (truncation of on kaci <
'won ka'ci ‘all kinds’), pam-nuc (= pam-nucengi) ‘chestnut blossoms’, pec (= pecci)
‘cherry’, pic < pit ‘debt’, pon koc’ (truncation of pon kocang) ‘native place’ (= pon kos <
104 PART I
A Reference Grammar of Korean
pwoti kwot).
ch coch- < cwoch- ‘follow’, ccoch- < pcwoch- ‘pursue’, kech = kecwuk ‘surface,
exterior’, kich- (? truncation) = kichi- ‘cough’, ich- ‘get tired’, mich- < mich- (? truncation of
michi-) ‘attain, reach’, nwiwuch- < "nwuyGuch- (truncation of nwiwuchi- < "nwuyGuchi-)
‘regret’, sich- (truncation of sichi-) ‘sew a quilt’, ssich- [dialect] = ssis- ‘wash’ < sis-,
takuch- (truncation of takuchi-) ‘bring nearer’;
'ach ‘reason’, 'cich > kis ‘feather(s)’, isulach = isulac(h)i (dialect) < i'su'lac
‘(wild) cherry’, kach < kach ‘skin, hide’, kkoch < (••• s) kwoc ‘flower’, meych (myech) <
'myech ‘how much/many’, mich [literary] ‘and’, nach < noch ‘face’, nuch < nuc [obsolete]
‘sign, portent, omen’, pichi < pyeth ‘sunshine’, pich 2 < pich ‘color; sign, mark; scene(ry)’,
'such ‘time interval, while’, swuch < swusk ‘charcoal’, tach (tech) ‘anchor’, tech (dialect
tek) ‘snare; small drum’ < *tesk, toch < twosk ‘sail’, yuch ?< *nywusk ‘Four Sticks (a
Korean game)’. NOTE: The noun och ‘sumac, lacquer’ was attested as woch in 1608 Thaysan
53a ( wo.ch ol) but earlier it appeared as wos in 1463 Pep 1:219a (wo's i'la).
ph ciph- < tiph- ‘lean (hands) on, feel (pulse)’, eph- < eph- ‘overthrow’ (rare except in
compounds and twicipeeph- ‘turn inside out or upside down’), iph- = ulph- > ulph- ‘chant’,
kaph- < kaph- ‘reward, repay’, kiph- < kiph- ‘be deep’, noph- < nwoph- ‘be high’, siph-
< sikpu- ‘be inclined toward, be desirous’, teph- < teph- ‘cover with, use as a cover’, tu-
noph- ‘be lofty’, thoph- ‘search everywhere for; soften and spread hemp tufts (to make rope)’,
? puph- (= puphu-1-) in puph-tay- and puph-ta-1- (but the aspiration could not be realized
here, so this seems to be a purely orthographic or historical example).
aph < a(l)ph ‘front’ (and compounds such as ocil-aph ‘front of an outer garment’),
ciph < 'tiph ‘straw’, hengkeph < heng'kes < "hen (< *he'[l-o]n) kes ‘piece of cloth’ [-ph
unexplained], iph < niph ‘leaf’, iph ‘gate’, keseph ‘a levee reinforcement; a weed pot-
cover; vegetables for pipimq pap’, muluph < mwulwuph ‘knee’, nuph (dialect nwuph)
‘marsh, swamp’, 'pwuph > puk ‘drum’, seph < syeph ‘kindling, firewood; gusset; prop’,
swuph < 'swuph ‘forest’, yeph (/••• nyekh) < nyephlnyekh ‘side, flank’,
ps eps- < "eps- (?< *e- pfijsfij- or *e'pV is[i]-) ‘be nonexistent’ and certain stems
derived from it: ka.yeps- ‘be pitiful’, mayk-eps- ‘be despondent’, sil-eps- ‘be frivolous,
unsubstantial’, silum-eps- ‘be absentminded, vacant’, sokcel-eps- ‘be futile, hopeless’, yel-eps-
‘be timid, cowardly’;
kaps < 'kaps ‘price’,
kh - ;
puekh < puzeklpuzep ?< *'pu[l] s(y)ep[h] ‘kitchen’, - nyekh ‘direction’ (see yeph).
kk < sk kkakk- < kosk- ‘cut, shave, pare’, kkekk- < kesk- ‘break off’, kyekk- < kyesk-
‘experience, undergo’, mukk- < mwusk- (?< *'mwufl] sk-) make into a bundle’, nakk- ‘fish’
from naks- < naksk- (Cf naks-i spelled nakk-si ‘fishing’; SEE -si in Part II), pokk- < pwosk-
‘roast’, sekk- < sesk- ‘mix it’, sokk- (dialect sokkwu-) < swos'kwo- (?= swos-'kwo- ‘raise’)
‘weed out’, takk- < task- ‘polish’, tekk- ‘get dirty/rusty’, yekk- < yesk- ‘knit, weave’;
pakk < pask ‘outside’; pusk (?< *pu[l] sk-) ‘moxacautery’, swusk > swuch
‘charcoal’, twosk > toch ‘sail’, twosk > tos(-cali) ‘mat’,
ks - ;
moks < mwok ‘portion’ (™s unexplained; blended with mwus ‘bundle’ < mwusk-1),
neks < neks ‘spirit’, saks < saks i ‘charge, fare’, sakso > ssak ‘sprout’, seks ‘surge of
emotion (especially anger); a mooring’, 'syeks ‘reins’, ches-paks [dialect] ‘first’, mayks
[dialect] = mayk ‘pulse’; naks ‘fishhook’ (> nakk.si).
h (For MK -h nouns, see p. 109.) ? ccah- [dialect] = cca- < pco- ‘weave’, ccih- < tih-
‘pound, ram’, ceh- ‘fear’- also (1465 Wen 1:2:3:40a) cyeh-, cih- ‘affix’, coh- < "tywoh-
‘be good/liked’, 'cwoh- ‘be clean’, eh- < fpjeh- ‘get cut’ (mistakenly treated as “eth-” in
A Reference Grammar of Korean
PART I 105
LCT 558a, the correct analysis is LCT 1971:22) - Cf e'hi- < [p]e'hi- ‘cut it’, nah- < nah-
‘be born’, neh- < nyeh- ‘put in’, noh- < nwoh- ‘put’, peh- ‘get cut’, pih- ‘sprinkle, sow’ (?<
pi ‘rain’ + °ho- ‘do’), ppah- ‘grind’, ssah- < (s)sah- ‘pile/heap up, build’, spih-lspyeh- =
pih-, tah- < tah-\ ‘touch; arrive’, ttah- < tah- 2 ‘braid’. (See §8.2.2.)
mh (For MK -mh nouns, see p. 109.)
nh (For MK ~nh nouns, see p. 109.) anh- (< a'ni ho-) ‘not do/be’ (negative auxiliary)
and compounds that contain it, hunh- < hun ha- < hun °ho- ‘be common, plentiful, easily
had, cheap’, kkonh- ‘mark, grade, rate’, kkunh- ‘break/cut off; stop’, manh- < "manh- <
"man °ho- ‘be much/many’. (See §8.2.2.)
lh (For MK -lh nouns, see below.) alh- < alh- ‘ail’, halh- > halth- ‘lick, taste’, helh-
= hel ha- < hel °ho- ‘be easy, undemanding’, ilh- < ilh- ‘lose’, kolh- < kwolh- ‘be
unfilled, half-empty’, kolh- < kwolh- ‘rot’, kkulh- < kulh- ‘boil’, kkwulh- ‘bend knees (to
kneel)’, olh- < wolh- ‘be right’, silh- < 'sulh- ‘be disliked’, ssulh- < sulh- ‘polish (grain)’,
talh- ‘wear away; boil dry’, ttwulh- < "tulw-(l tolw-) ‘pierce’. (See §8.2.2.)
Is - ;
kols ‘(water-)course, (fixed) direction’, "kwols > kol(ay) ‘hypocaust flue’, ols
‘compensation, reparation’, tols < twols ‘cycle, postnatal year of age’.
1th halth- < halh- ‘lick, taste’, hwulth- ‘tear off something stuck to the surface, rinse out
something stuck inside a bowl; thresh’;
lk elk- < elk- ‘wrap, tie up, fasten’, kalk- < kolk- ‘scratch with a sharp point’, kulk- <
kulk- ‘scratch’, kwulk- < kwulk- ‘be burly’, malk- < molk- ‘be clear’, mulk- < mulk- ‘be
thin, watery’, nalk- < nolk- ‘(thing) be old’, nulk- < nulk- ‘(person) get/be old’, oik- ‘trap,
ensnare’, palk- < polk- ‘dawn; get/be bright’, (p)pulk- < pulk- ‘be red’;
chilk < chulk ‘arrowroot; striped’, hoik (variant of hoi) ‘a growth’, hulk < hoik
‘earth, soil’, katalk (= katak) ‘strip, piece, strand’, kkatalk ‘reason’, kisulk (= kisulak)
‘edge, border’, selk < selk [obsolete] = selki ‘wicker trunk’, salk < solk ‘leopard cat’,
siwulk (variant siwul) < si'Gwulk ‘edge’ [old-fashioned] (= kacang-cali), talk < tolk
‘chicken’. *Yi Yuncay also gives the pre-separable intransitive verbal noun inseparable wulk
(ha- ‘get rash/hasty’) but I am unable to find evidence that the 1 is ever pronounced; the
spelling may be historical or based on an association, morphemic or dialectal, with the stem
oik- or with the stem pulk-. ('Yi Yuncay was mistaken in labeling the word adjectival.)
/G//% < *l u /ok See §3.3.1.
////%/ < */%/ See §3.3.1.
lm celm- < cyelm- (1775) < "cyem- ‘be young’, cilm- ‘bundle up to carry, pack on
back’, "kalm- ‘store/hide it’, kolm- < ["jkwolm- ‘fester’, kwulm- < [ Jkwulm- ‘go without
food’, palm- ‘measure off by the arms; guess’, salm- < "solm- ‘boil’, talm- ‘resemble’ <
"talm- ‘spread (disease)’, "telm- ‘get dirty/dyed’, olm- < "wolm- ‘move’ (= olm.ki-) and ‘be
infected by, catch (a disease)’;
salm ‘life’, aim ‘knowledge’, and all other regular substantives from the 1-extending
vowel stems (§8.2.2, §9.5)
lp ccalp-/ccelp- (< cyelp-) < tyalo- / tyel %- < tyelG- < *tyelW-l< *tyalop-1tyelup-
‘be short, fine’, "kolW- ‘line up, array; compete’, nelp- < nel %- ?< *nelup- ‘be wide’,
palp- < "polW- ‘tread on’, selp- (= selw- [obsolete] = selew-) < "syelW- ‘be sad’, ttelp- <
"ptelW- ‘be astringent’, yalp-/ yelp- < "yelW- ‘be thin, faint, light’;
'kolp ‘layer; time’, 'salp > sap ‘shovel’, yetelp (dialect yatul, yatap, yatak) <
ye'tulplye'tolp (< *yotolp) ‘eight’ (see below).
lph ulph- < ulph- / iph- ‘chant’; aytalph- < "ay tolp- ‘feel pity’, kotalph- [lit.] =
kotalphu- ‘be tired’ (< kwo'tol'ph-a ‘with great effort’);
- , alph = aph > aph ‘front’.
nc anc- < a(n)c- ‘sit down’, enc- < enc- < yenc- ‘place, put up/on’, kki-enc- ‘shower
106 PART I
A Reference Grammar of Korean
oneself’;
nk/nu < *nuk See §8.3.1.
mklmo < *mok See §8.3.1.
mklmwo < *mwok See §8.3.1.
mklmwu < *mwuk See §8.3.1.
zG/z% < *s u fok See §8.3.1.
Not included in the list are the names for letters of the Korean alphabet. These are rather artificial
concoctions, usually pronounced according to a common variant: tikut/tikus ‘the letter T’, thiuth
/thius ‘the letter TH’, chiuch/chius ‘the letter CH’, phiuph/phiup ‘the letter PH’, khiukh/khiuk ‘the
letter kh’. There is also hiuh, the only known case of noun-final h in the standard orthography. The h
is treated as Itl before pause or consonant, and should be either /h/ or dropped before the copula or a
particle beginning with a vowel, but in fact this word nearly always gets the variant treatment
represented by the basic shape hius. The bound preparticle pa.yah (§5.2.9) is written together with its
particle ulo as an unanalyzed word: pa-ya-hu-lo. The verb stems with final -lph- carry a literary
flavor; they seem to be truncated from vowel stems ending in -lphu-, as shown by the third example.
An example of -nth is found in Khun sacen, which lists panth as a variant of pan ‘half’. I am
told that /panthun/ is South Cenla dialect for pan un ‘as for half’. 1 Yi Yuncay lists panth as
Kyengsang dialect. The th is etymologically unexpected and its origin is unknown (?< hatun).
When followed by a vowel that is NOT the beginning of an ending, a particle, or the copula,
syllable excess is reduced just as before a consonant, so that the common noun-final -s is pronounced
t and articulated as the onset of the vowel-initial syllable to which it is attached. Examples:
s -*■ t: os an ‘inside the garment’ /otan/, kulus an ‘in the plate’ /kulutan/, ches atul ‘first son’
/chetatul/, ches insang ‘first impression’ /chetinsang/, ches umcel ‘first syllable’ /chetumcel/, kis os
‘a kind of mourning robe’ /kitot/, has os ‘padded garment’ /hatot/, wus os ‘outer garment’ /wutot/,
wus akwi ‘crotch between thumb and index finger’ /wutakwi/, Oypus atul (emi, epi) ‘step-son
(-mother, -father)’ /uputatul/ (/Oputemi/, /uputepi/), yeys wang ‘ancient kings’ /yeytwang/, swus
umsik ‘fresh food’ /swutOmsik/, hes wus.um ‘empty smile’ /hetwusum/, pelus eps- ‘lacks manners’
/pelutepss-/, ... .
c t: cec emeni ‘wet-nurse’ /cetemeni/, .... Compare cec hyeng ‘older nursemate (“milk-brother”)
/cethyeng/.
ch -» t: kkoch ahop songi ‘nine flowers’ /kkota(h)opssongi/, kkoch alay ‘under the flower’
/kkotalay/, hayq-pich ani ’myen ‘unless it is sunshine’ /hayppitanimyen/, och olu- ‘get lacquer-
poisoned’ /otolu-/.
t: mat atul ‘eldest son’ (but mat ita ‘is the eldest’ Seoul /macita/), ... . Compare mat hyeng ‘the
eldest brother’ /mathyeng/.
th -* t: soth an ‘in the pot’ /sotan/, path alay ‘below the field’ /patalay/, path wi ‘above the
field’ /patwi/ (compare path twl ‘behind the field’ /pattwl/), hoth os ‘single-layer garment’ /hotot/,
kkuth ani ’ta ‘it is not the end’ /kkutanita/, pith eps- ‘lack color’ /pitepss-/, puth-an- ‘hug’ /putan-/,
sath-sath-i ‘in every corner, exhaustively’ /sassachi/.
ps -*■ p: kaps echi ‘worth’ /kapechi/, kaps eps- ‘lack value’ /kapepss-/ (compare kaps ci- ‘be of
value’ /kapcci-/).
ph -*■ p: aph.aph-i ‘in front of each’ /apaphi/, iph wi ‘on the leaf’ /ip(w)i/ (Cf §2.7.4), muluph
wi /mulup(w)i/ ‘on the knee’. Compare noph-tala(h)- ‘be sort of tall’ /nopttala(h)-/.
kh -* k: puekh an ‘in the kitchen’ /puekan/.
ks -*• k: neks eps.i ‘absentmindedly’ /nekepssi/.
Is -» 1: tols an ey ‘within the first year of life’ /tolaney/.
The following cases involve reinforcement (-q, Cf §1.5.): alayq i ‘lower teeth’ /alaynni/, aphq
A Reference Grammar of Korean
PART I 107
ima ‘forehead’ /amnima/, cipq Imca ‘householder’ /cimnlmca/, hothq ipul ‘single quilt’ /honnipul/,
kyepq ipul ‘double quilt’ /kyemnipul/, pamq isul ‘evening dew’ /pamnisul/, sokq iph ‘the inside
leaf’ /songnip/, ttekq iph ‘seedleaf’ /ttengnip/, wiq ip-swul ‘upper lip’ /winnipsswul/ ‘upper lip’.
For certain examples the reinforcement may be optional: pathq ilang ‘field ridge’ /pannilang/ is also
reported as path ilang /patilang/. The orthographically identical path ilang ‘field and the like’ with
the colloquial particle ilang/lang is pronounced /pachilang/, and many speakers say /pasilang/.
In a few expressions, contrary to what we expect, the syllable excess persists, as in yetelp hay
‘eight years’ /yetelphay/ where we would expect (?*)yetel(h)ay. The word for ‘eight’ has been
restructured as yetel(q) for most speakers, who say yetel ita and not yetelp ita for ‘they are ten’. In
causatives and passives, the derived stems preserve as much of the excess as can be pronounced: palp-
hi- ‘get trodden on’ /palphi-/ is the passive of palp- ‘tread on’, but kolm-ki- ‘has it fester’, the
causative of kolm- ‘fester’, must be reduced to /komki-/. And most derivatives are like compounds
and reduce the excess: noph-talah- ‘be sort of tall’ /nopttala(h)-/, nelp-tala(h)- /nelttala(h)-/ ‘be sort
of wide’, kwulk-tala(h)- ‘be sort of burly’ /kwukttala(h)-/ or /kwulttala(h)-/.
'Yi Ungpayk 454 calls our attention to the rule by which compounds are spelled phonemically
rather than morphophonemically if the last consonant of a double-consonant pat.chim is not
pronounced:
kolmak / kwulmek ha- ‘be almost full’ from kolh-/kwulh- ‘be not yet full’ (but notice kwulm- ‘go
without food, starve’);
kolpyeng ‘deep-seated disease; fatal blow’ from kolh- ‘rot’ + pyeng ‘illness’;
halccak halccak ‘in little licks’ from halth- ‘lick’ + -cak (diminutive suffix < cak- ‘little’);
silccwuk/saylccwuk ha- ‘be sullen’ from silh- + suffix -cwuk;
malsswuk/melsswuk ha- ‘be neat’ from malk- ‘clear’ + suffix -swuk;
malccang/melcceng ha- ‘be intact, perfect’ from malk- ‘clear’ + suffix -cang/-ceng;
olmu ‘snare’ from oik- ‘bind, lay a snare’ + ?; one proposed etymology has olk.a kam- > *olk.a
kam-i > olkami ‘snare’ > *olkamu > *olk’mu > olmu. It is unclear just why silh-cung (=
silhq-cung) ‘ennui’ is not spelled according to its pronunciation /silccung/.
Moreover, there are examples of phonemic spelling even when the syllable excess is pronounced:
yalphak ha- ‘be thin-surfaced’ from yalph- ‘be thin’ (with syllable excess retained) + suffix -ak, or
(with syllable excess suppressed) + the mimetic phak ‘deflated, flat; soft’, probably unrelated to
Chinese loanmorph pak < PPAK ‘thin’;
silkhum ha- ‘be dislikable’ from silh- + suffix -kum (Cf silh.ko ‘disliking’ /silkho/).
The structure of malkkum / melkkum ha- ‘be clean’ can be explained as a reduction of malk- -» mal-
before attaching the suffix, which then reinforces its initial, as it does in malk-ko and malk-ci when
pronounced /malkko/ and /malcci/ rather than the competing version /makko/ •*- mak-ko and
/makcci/ *- mak-ci.
Some confusion exists over whether there are two versions of ‘rather wide’: /nepttala(h)-/ =
nelp.talah- and /nelttala(h)-/ = nelttala(h)- (spelled nel.tala(h)- in North Korea). NKd lists both
versions and suggests that the latter comes from nelu- ‘be broad’; most South Korean authorities prefer
the second version (nel— not nep-) but derive it from nelp-. A similar problem: /nepccik/ and
/nelccik/. The South Korean authorities seem to prefer the latter and spell it phonemically. NKd lists
nelp.dk (presumably to be pronounced /nepcik/) but refers it to the entry nel.dk for which the
pronunciation is explicitly stated as /nelcdk/; there are entries for /nepccek/ spelled both nelp.cek and
nep.cek, and similarly for /nepccwuk/.
Finally, we should keep in mind the ongoing tendency to restructure the basic forms of most
nouns bearing syllable excess. Even Seoul speakers often simplify noun-final -ps to just -p; the
pronunciation /kap/ for kaps ‘price’ in umsik kap[s] ey nun in drill 3.7 of KBC 24 is not a slip of the
tape or the tongue, but a variant of what is heard as /kapss/, a more formal version, in mulken kaps
ey nun in drill 3.9. In everyday speech people quite often say /kapun/ and /kapulo/ (for kaps un,
kaps ulo) instead of /kapssun/ and /kapssulo/, even though they may well say /kapssi, kapssita/ for
108 PART I
A Reference Grammar of Korean
kaps i, kaps ita. The liquid in talk ‘chicken’ is widely ignored, so that tak i and tak ul are the
commonly heard nominative and accusative forms; the liquid is retained in a derivative talkyal ‘egg’.
In modern usage moks ‘portion’ and kols ‘course, channel’ are generally just mok and kol: Ku ttang
un nay mok[s] ip.nita; ney mok[s] un ... ‘That land is my share; your share ... ’; Kol[s] ul kiph.key
pha la ‘Dig the channel deep!’. We can treat puekh ‘kitchen’ as an obsolescent version of the widely
used puek (also pek); but pakk ‘outside’ persists unsimplified. The noun muluph ‘knees’ is heard
either with the simplified basic form mulup or in a derivative muluphak(-i), which preserved the
syllable excess when the diminutive suffix -ak was added. But aph ‘front’, yeph ‘beside’, and iph
‘leaf (also iphali, ipsakwi) commonly retain the older basic shapes. Modern Seoul speech is in flux
on the question of merging noun-final — th, ••• ch, and — c with — s. The four-stick game is generally
treated as yus rather than the older yuch; och ‘lacquer; sumac’ is treated just like os ‘garment’; and
path is more often heard with final s than with the traditional th or (before i) ch. But both meych and
meys are common for the orthographic myech ‘how many’, and only the affricate is heard in nac ey
‘in the daytime/afternoon’, though you may notice an allophone of that which is articulated as a
voiced sibilant [z]. There are no good examples of noun-final -t, since nat of nat-al ‘grain’ is no
longer a free noun, and earlier cases of noun-final -t merged with ™s over two centuries ago, so that
'mwot ‘nail’ and mwos ‘pond’ are now both mos. Despite these remarks on colloquial usage, this
book follows the traditional spellings and we treat syllable excess as basic to those nouns that began
simplifying it relatively late.
Earlier forms of the language had a wider array of extrasyllabic finals, including stems ending in
-sk- (> -kk-) and -st- (> -th-). There were even a few nouns without vowels: see in Part II the
entries psk ‘time’, s ‘fact’, t ‘fact’. The reduction of the excess was similar to what it is today, but in
the 15th century there was no need to reduce -s since it was a syllable-final consonant. Stems ending
in ~(n)c- such as a(n)c- ‘sit’ reduced to ■•■£-, as did those ending in -z- (lenited from ■■■$■■■) and those
ending in -sk- and -st-. The doublet wumchu- / wumchi- ‘huddle, shrink’ contracted to a reduced stem
wums- in a few examples. The noun poych ‘oar’ must be the result of truncating an unattested phrase
'poy chi ‘boat rudder’.
In the modern language final -h, -nh, and -lh occur only for verb stems, but in Middle Korean
there were nouns that had these basic codas (and also -mb), which surface as the aspirating of a
voiceless consonant that begins a following particle, though the /h/ was suppressed when the noun was
in isolation, i.e. before pause. As mentioned above, there is evidence for some of these noun-final h’s
in such modern compounds as am-khay < amh kay ‘female dog’ and swu-khay ‘male dog’ < swuh
kay. The spellings of 1898 Tayshin show final h for a number of nouns with the particle i (or
incorporated -i) attached: ttahi, tta ‘earth’; patahi, pata ‘sea’, twuihyi ‘behind’ (= Seoul twl ey); and
even one case of final h in isolation, narahi, narah. (1874 Putsillo gives ‘behind’ as twui, twuhe.) All
these nouns had a basic final -h in Middle Korean. The standard noun nai ‘age’ must have developed
like the Tayshin nahi, by incorporating -i to the MK noun that had the basic shape nah and survived
in its “free” shape as na in modern dialects (e.g. the South Hamkyeng version used by 1936 Roth
197); Cf naq sal ‘age, years [often derogatory]’ (KEd). 1894 Gale writes (48) hon.a.hun ‘as for one’
(< MK hona'hun), and (148) hon.a (< MK hona[h]) with an unexpected syllable break perhaps
reflecting the allomorph hon- but missing in hona.hi (80); also (82) pata.hi and pata.hul for the
nominative and accusative of ‘sea’ (< MK pa'tah), and (64) chol.ha.li for chalali ‘rather’ (< MK
chol'hali).
Some of the stems ending in the final aspirate seem to have incorporated the pro-verb °ho- > ha-
‘do/be’, which was prone to elide its minimal vowel, leaving the h behind to fend for itself. Although
ilh- is the only MK version for ‘lose’, 1898 Tayshin attests both ilhata (with palatalized [1]!) and iltha
(“irta” with [r]!). *Yi Congchel 1983 clearly writes man ha- (with the Chinese character MYRIAD) for
manh-; the basic stem in the earliest Hankul texts is sometimes "man °ho-, sometimes "manh-. The
basic form of MK word for ‘above’ was wuh; the final h was suppressed when the noun preceded
juncture (that is, when it was not followed by a closely attached particle or copula form), and it is that
form which survives as the standard North Korean wu, while Seoul has standardized wi < wu[h] +
A Reference Grammar of Korean
PART I 109
incorporated -i, a development more common as a northern characteristic. (1894 Gale 133 writes mwul
wu.huy ‘in the water’.) NKd lists a dialect form wuthi, not in the Hamkyeng dictionary nor in Choy
Hak.kun (though he has “uge” = wukey), which may reflect the earlier h, but the derivation is
unclear. There are two examples of ~ngh: stang.h ay (1617 Sin-Sok 'Yel 4:64) ‘on the land’ seems to
be a hybrid of earlier sta'h ay and stang (id. Hyo 1:1), but syang'h ay ‘regularly’ (1518 Sohak-cho
8:9b) is not so easily explained.
In the following lists of MK nouns ending in -h a few of the examples occur as doublets, with or
without the h in the basic form. That means there are competing phrases without the h where we would
expect it. (Of course, all the nouns suppress the h when they occur as free forms.) The two versions of
the doublets are separated by a slash.
MK nouns with final -mh: 'amh ‘female’, 'wumh ‘cave’.
MK nouns with final -nh: 'anh ‘inside’; enh ‘dike’; kinh, 'skinh ‘string’; twuy-anh (1576 'Yuhap
2:28b, 1632 Twusi-cwung 6:50a) = wuy-'anh ‘garden’.
MK nouns with final ~lh: alh ‘egg’ 'cholh ‘source’ (Cf ' stolh ); ha'nolhlha'nol ‘heaven’; 'kalhl
'kal ‘knife’ ('kal 'Iwo 1465 Wen 3:2:2:10a = 'kal.h o['Jlwo 1462 , Nung6:109b; 'kal Gwa 1462 'Nung
6:28b = kal'khwa 1459 Wei 9:43b); kilh ‘path’; 'ko(')nolhl ko(')nol ‘shade’; kowolh ‘district’ (=
kowolkh, koWol); kozolhlkozol ‘autumn’; kye'zulhlkye'zul > kyefzjolfh) winter; 'malh ‘stake’; milh
‘wheat’; mozolh ‘village’; 'nolhi ‘blade’; 'nolhi ‘warp (threads)’; nomolh ‘greens’; polh > pholh >
phol ‘arm’; poyzolh/poyzol (< *poy solh) ‘entrails’ (poyzol tol'h ol 1463 Pep 2:105b); 'ptulhl 'ptul
‘garden’ ('ptul 'Gwa Twusi 25:39a); pyelh ‘cliff’; 'solh ‘flesh’; 'stolh ‘origin, source’ (Cf 'cholh);
'sukwulh ‘rural area’ (= 'sukwol, 'sukowol, 'sukoWol); su'mulh ‘twenty’; "syewulhl "syewul,
"syeWul[h] ‘capital’ ("syewul 'Iwo 1481 Twusi 24:45b, "sye'wul 'set'two ? 1517- Pak 1:53b [accent
unexplained]); tolh ‘group’ (plural); "twolh\ ‘bridge’; "twolhi ‘stone’; "twulh ‘two’; wolh ‘this year’;
'wulh ‘fence’; yelh ‘ten’; yelh ‘hemp seed’.
MK nouns with postvocalic final -h: 'cah ‘foot’; cwoh ‘millet’; honah ‘one’; kuluh ‘root’; 'kwoh
‘nose’; mah ‘yam’; mah/ma ‘monsoon’ (= tyang-ma[h]), ‘south, south wind (= *mahpolom > ma-
phalam); 'moyh ‘moor’; "mwoyh mountain’; 'mwoh ‘corner’; na'cwoh ‘evening’ = na'cwoy ; nah
‘age’; na'lah ‘land’; "nayh ‘river’; "neyh ‘four’; ni'mah ‘forehead’; "nimcahl "nimca ‘master; you’;
nwoh ‘cord’; pa'tah ‘sea’; pwoh\ ‘beam’; pwoh 2 ‘cloth’; "seyh ‘three’; stah ‘ground’; tuluh ‘hat brim’
( ? 1517- 'No 2:52a); tu luh ‘moor’; "twuyh ‘rear’; tyeh ‘whistle’; swoh^ ‘swamp’; 'swoh 2 ‘matrix,
mold, die’; swuh^ ‘male’; swuh 2 ‘forest’ = 'swuph; sywoh ‘layman’; "si[lj-"nayh/-"nay ("si"nay
Iwo 1481 Twusi 21:34a) ‘stream’; 'theh ‘site’; wuh ‘above’; ye'leh ‘several’; zywoh ‘mattress’.
4.3. Treatment of yey.
The phoneme string /yey/ occurs only after a pause; in other positions it is automatically replaced
by ley I, so that pon yeysan ‘the main/original budget’ is pronounced either /ponlyeysan/ or
/poneysan/. The string leyl itself begins the basic form of very few words (eywu- ‘surround’, eyi-
‘cut’, ... , and recent loanwords) so that it is infrequent after pause.
Among the Chinese morphemes, the South Korean spelling writes phyey for /phey/ as in ph y ey
‘lungs’, and myey for Imeyl in the bound noun m y ey, which appears in uym y ey ‘sleeve’, m y eykwu
‘sleeve opening’, *yenm y ey ‘(in) company’, and punm y ey ‘parting (of people)’. Both North and South
spell hyey for /hey/ in Chinese loans such as hyeyseng ‘comet’. Certain elements beginning with /key/
are distinguished in Hankul by the spelling kyey, such as kyeysi- ‘stay [honorific]’, kyeysi ‘revelation’
(pronounced just the same as keysi ‘notice, bulletin’), but there seems to be no good reason for any of
these spellings, except perhaps historical. The spelling conventions are that only South Korea writes
the unpronounced distinction of myey : mey and phyey : phey, but both South and North write the
unpronounced distinctions of kyey : key, hyey : hey, and yey : ey. (Cf Mkk 1960:9:37-8.)
Because of the automatic alternation of yey with ey, morphemes with the basic shape lyey (such as
the common one meaning ‘rite, ceremony’) never actually occur in that shape at all; it is a Active form
based on the occurring alternants /ley/ (as in sillyey ‘discourtesy’ /silley/ and kolyey ‘ancient rites’
110 PART I
A Reference Grammar of Korean
/koley/) and /yey/, as in 'yeypay ‘worship’.
In addition to word variants like /••• eysan/ for yeysan, which have to be caught on the fly and can
be ignored for most purposes, we also observe the alternation of ley / and lyeyl in morphemes with the
basic shape yey, such as the one that means ‘esthetic, art’ and occurs initially in yeyswul ‘esthetic
techniques’ and finally in mun.yey ‘literature and art, humanities’ = /muney/, hak.yey ‘science and
art = arts and sciences’ = /hakey/, kongyey ‘arts and crafts’ = /kongey/, kiyey ‘crafts’ = /kiey/,
swuyey ‘handicraft’ = /swuey/, etc. The dialect variants ney (northern and modern-Seoul) and yey
‘yes’ can be seen as a somewhat similar case, from a basic *nyey.
4.4. Alternations of 1 and n.
Except in recent loanwords (latio ‘radio’, nyusu ‘news’, nikheyl ‘nickel’), in a few native
oddities (see below), and in the grammarians’ neologisms liul ‘the letter l’ and niun ‘the letter N’, the
phoneme 1 does not occur after pause, nor do the strings ny and ni. In older loanwords 1- -* n-:
‘Nwuka pok.um ‘the gospel of Luke’, nampho(-tung) ‘lamp’. After pause, a morpheme whose basic
allomorph begins with 1 appears in an alternant beginning with n. But those morphemes whose basic
allomorphs begin with li, ni, ly, or ny appear in allomorphs which begin with i or y:
BASIC SHAPE
SHAPE AFTER PAUSE
1 :
n
yolo ‘major road’
'nopyen ‘roadside’
n :
n
sinay ‘city confines’
naypu ‘inside part’
ly :
y
nolyek ‘effort’
'yek.hayng ‘exertion’
ny :
y
swunye ‘nun’
n yeca ‘woman’
li :
i
sail ‘reason’
*iyu ‘reason’
ni ;
i
*noni ‘a kind of clay’
n iyok ‘mud bath’
So far as the alternants after pause occur only after pause, they can be called phonemically
determined, provided we ignore the recent loanwords and a few native oddities such as nyesek ‘rascal
of a man’, nyen ‘rascal of a woman’, nyen-nom ‘men and women rascals’, niun ‘the letter N’. But in
most words the “altered” allomorph occurs word-initially whether the word is preceded by pause or
not: ku n yeca ‘that woman’, i 'nopyen, ... . That is sometimes obscured by the “-q” phenomena
discussed below.
Certain other cases must be specified in detail. The word for ‘league’ or ‘Korean mile’ has the
shape lil except after a numeral, where it has the shapes /li/ or /ni/ (written li): il li ‘one mile’, i li
‘two miles’, sam li /samni/ ‘three miles’, etc. The Chinese word for ‘two’ always has the shape IV
except after the word il ‘one’: il-i ‘one or two’ is usually pronounced /illi/. The word for ‘reason’ has
the shape IV (written *1) except after the prospective modifier: -ul li eps- ‘not stand to reason that - ’.
The Chinese word for ‘six’ has the shape lyukl (spelled *yuk) except after a numeral: sip *yuk = sip-
lyuk /simnyuk/ ‘16’, 'yuk-sipq 'yuk /yukssimnyuk/. See §5.5.
A number of words beginning with i- or yhave alternants beginning with ni- or ny- (or
reflexes of those strings) which appear in certain environments; these are best treated as cases of
reinforcement (-q): cipq II /cimnll/ or cip II /cipll/ ‘housework’, halq II /hallil/ ‘things to be done’. In
the case of the noun II the MK form had an oral beginning, but certain other nouns that nowadays
behave the same way were spelled as ni- or nye —: i ‘tooth’ was MK ni and iph ‘leaf was niph (in
contrast with iph ‘gate’) but ip ‘mouth’ was MK ip. We are tempted to write “alayq ni” for ‘lower
teeth’, and that would be historically correct, but we have no way of keeping that situation apart from
/alaynnipsswul/ ‘lower lip’ where “alayq nip-swul” would be historically incorrect. For the modern
language we will treat all cases alike and write not only alayq ip-swul but also alayq i, letting a rule
interpret “q i” as /nni/ and “q y” as /nny/ in examples such as hanq yeph = /hannyeph/ ‘one side’.
For Chinese words, the historical ny-, ly-, and li- are in general so written in the North Korean
orthography, and the initial nasal is preserved in the spoken dialects, with loss of the -y- (except in
’Yukcin). The initials of those strings are represented by superscript n and 1 in the Yale Romanization:
A Reference Grammar of Korean
PARTI 111
ettenq "yeca /ettennyeca/ ‘what sort of woman’, ponq "yento /ponnyento/ ‘this year period’,
musunq *iyu lo /musunniyulo/ ‘for what reason’, kulenq ^ek.hayng /kulennyekhayng/ ‘such
exertion’; yang 1 yoli /yangnyoli/ ‘western food’, Cwungkwukq 1 yoli /cwungkwungnyoli/ ‘Chinese
food’; sampaykq ^uksip-o il /sampayngnyukssipoil/ ‘365 days’. The behavior of these is not
distinguishable from that of the historically correct y~ and i-: musunq yoil /musunnyoil/ ‘what
day of the week’, nolanq yangmal /nolannyangmal/ ‘yellow stockings’ (KBC), kulenq ilmyen
/kulennilmyen/ ‘such a side (to it/him)’, Pusanq yek /pusannyek/ ‘Pusan station’, Sewulq yek
/sewullyek/ ‘Seoul station’, slcheng-aphq yek /sichengamnyek/ ‘City Hall Station’; sonnimq-yong
siktang /sonnimnyonglsikttang/(KBC2:24) ‘guest dining room’, chaykq yenkwu /chayngnyengkwu/
‘book research’. For certain strings the reinforcement is optional: kkoch(q) ilum ‘fewer name’ can be
said either /kkotilum/ or /kkonnilum/. The poetic noun im ‘beloved’ (so spelled in the north as well
as the south) was earlier nim and is probably the same morpheme as the honorific postnoun - nim <
MK "nim ‘esteemed ••• ’; the noun Imkum < MK "nim-'kum ‘king’ probably contains the same etymon.
Almost all verbs beginning with i- or y- have the reinforced form, but only after prefixes or the
negative adverbs mos and an: mosq ic.e /monnice/ ‘can’t forget’, mosq il.e na /monnilena/ ‘can’t
arise’, mosq ilk.e /monnilke/ ‘can’t read’, mosq ik.e /monnike/ ‘can’t ripen’, mosq ip.e /monnipe/
‘can’t wear’, mosq yel.e /monnyele/ ‘can’t open’; anq ic.e /annice/ ‘doesn’t forget’, anq il.e na
/anniiena/ ‘doesn’t arise’, anq ilk.e /annilke/ ‘doesn’t read’, anq ik.e /annike/ ‘doesn’t ripen’, anq
ip.e /annipe/ ‘doesn’t wear’, anq yel.e /annyele/ ‘doesn’t open’. For the stem iss- ‘stay, be’ the q
seems to be optional for mos but usual for an (because of pervasive n-epenthesis): mos(q) iss.keyss.e
/motikkeysse/ or /monnikkeysse/ ‘can’t stay’, anq iss.keyss.e /annikkeysse/ ‘won’t stay’. The Middle
Korean source of iss- was spelled without an initial nasal, isi-, but almost all the other relevant stems
were spelled ni- or ny-. For those verbs the non-reinforced treatment is sometimes heard (mos ic.e
/motice/, mos yel.e /motyele/), but not commonly. An example with a prefix is cisq-iki- /cinniki-/
‘knead; mince’. When the accusative particle is omitted in the phrase os ul ipko ‘wearing clothes’ the
phrase can be pronounced either /otipkko/ = os ipko or /onnipkko/ = osq ipko. If only the juncture
after the accusative particle is dropped, you may hear both /osulipkko/ with the flap [r] and
/osullipkko/ = os ulq ipko.
There are also words which begin with lyl but have alternants beginning with /ny/ in certain
environments. The word "yen ‘year’ after a numeral is pronounced /nyen/ (and that is automatically
/lyen/ after 1); the same pronunciation is common after ku ‘that’ and similar adnouns. The MORPHEME
for ‘year’ has the shape /yen/ spelled "yen when word-initial, but elsewhere it is nyen (including
/lyen/ after 1) elsewhere: 'naynyen ‘next year’; mal.nyen ‘the later (closing) years’; "yenkam
‘yearbook’; "yento ‘year period’; "yen-nyen ‘year after year’, Cf 'yennyen ‘successive years’ and
yennyen ‘prolonging one’s years (= life)’.
For a few words, such as those cited on p. 41, history has gone awry and confusion is rife. Some
Koreans treat -n-n- the same as -n-1- and —1-1—, namely as IWI, and say Allyeng hasimnikka for
Annyeng hasip.nikka ‘How are you?’. The words kwannyem ‘idea’ and konnan ‘difficulty’ are often
treated as if kwan.lyem /kwallyem/ and kon.lan /kollan/, and those spellings are included in some of
the dictionaries, and we have taken account of a few of these by writing such Romanized versions as
kon.fean. (The word /kwellyen/ ‘cigarette’ is usually spelled phonemically, though etymologically it is
kwen.yen.) Somewhat similar cases are kilyem for kinyem ‘memory, souvenir’ and kilung for klnung
‘talent, ability’. Double /II/ sometimes appears for no good reason where a single l\l is expected; such
forms are usually to be regarded as dialect variants. Occasionally, reinforcement (-q) is involved:
/mullyak/ ‘liquid medicine’ is best treated as mulq yak. That is perhaps the best way, too, to treat
‘one or two’: ilq-i.
Dialects of the northwest dispalatalized the older initial ty thy ny, while the southern dialects
affricated t(h)y so that they merged with the affricates c(h)y and retained the glide but dropped the
nasal of ny. (Only 'Yukcin preserves the original situation.) That is why for Seoul "yeca ‘woman’ and
chengke-cang ‘station’ (< tyengke-tyang ) northerners are known to say neca [nsdza] and tengke-tang.
112 PARTI
A Reference Grammar of Korean
These phenomena were noticed by 'Yu Huy in 1824 Enmun-ci (p. 7): “In Korean pronunciation tya,
tye have become cya, eye; thya, thye have become chya, chye. Only in Phyengan province do people
not equate thyen ‘heaven’ with chyen ‘thousand’ and ti ‘earth’ with ci ‘arrival’.” There are examples
of c for ty and ch for thy in the 1632 edition of Twusi enhay (Cf An Pyenghuy 1957). Seoul
irregularly has neh- < nyeh- ‘put in’ where we would have expected yeh- (as in various dialects),
reflecting the MK variant neh- attested in 1466 Kup (see p. 47) and perhaps influenced by the initial of
noh- ‘put’. 1898 Tayshin writes nyetta (?= /nyetha/) ‘lay, stow’. (Tayshin nipsiuely ‘lips’ must be a
back formation, for MK had ip si'Gwul without the nasal initial.) The northern tendency to
dispalatalize has weakened in the 20th century, and the southern palatalizations in loanwords such as
latio > lacio ‘radio’ and tisuthoma > cisuthoma ‘distoma’ are now common in the north, as well
(Mkk 1960:9:39).
For more on ny- and ni- see §1.2. For further discussion of alternations involving 1 see §5.1,
§8.1.4, §§8.2.1-3; §4.7, §4.8; §1.6; §2.6. Cf. Thak Huyswu 1956:160-7.
4.5. Shape types of Chinese vocabulary.
The table on pp. 114-5 shows all the shape types that occur in “normal” basic readings of Chinese
characters. With the exception of those queried with question marks, which were included out of
deference to Korean dictionaries, I believe examples can be found of real words containing morphemes
with shapes to justify the inclusion of each of the entries in the table. Distinctions of vowel length are
ignored. Certain shapes (such as lye) are always short, regardless of the particular morpheme
represented, and certain shapes are always basically long (like the, which represents a single
morpheme). But other shapes are distinctively short or long depending on the morpheme. The long
vowel in these shapes usually corresponds to the Middle Chinese “rising” and “falling” tones, but
there are many exceptions. Parentheses enclose marginal or special shapes.
Although there are characters that are to be read with the syllables cum, cwul, nin, nwal, nyek,
nyep, and phik, they are not used in loanwords found in modern Korean, so those have been left blank
in the table. There are no characters read with the syllables ewang, hi (as distinct from huy), kul, kya,
mam, non, nyak, nyang, nyey, op, phyu, or pik; these, too, are blank in the table. Among the filled
slots in the table, several shapes appear in only a few loanwords and some of the n- shapes do not
appear in environments critical for deciding the initial. For kh the only shape is khway.
A few of the rarer shapes, with examples:
hyul kwuhyul ‘relief (of the poor)’, hyulkum ‘relief fund’
kyak kyak.kum ‘collecting funds’
nun n imkum ‘pay’, n imtay ‘lease’
nwul nwul.en ‘stammering speech’, mok.nwul ‘innocence and lack of eloquence’;
but enwul ‘inarticulate’ is commonly said as /elwul/
nwun nwun.cho ‘fresh grass’, nwunhan ‘mild cold’, nwun.lok ‘light green’
nyel "Yelpan ‘Nirvana’
nyuk n yuk.hyel ‘nosebleed’
phyak phyak hata ‘is snippy’, koyphyak hata ‘is fussy’
thum thum.ip ‘trespassing’
The shapes kkik, ssang, and ssi are anomalous in beginning with reinforced consonants, but the
North Koreans standardize kkik ‘ingest’ as kik, and ssi was spelled si in earlier times. The reading
ssang ‘pair’ first appeared in 1677, but the reading ssi ‘clan’ seems to be fairly new, and is probably
the result of truncation from compounds (™q si ‘the clan of ’), just as the initial reinforcement of a
few nouns such as kkoch ‘flower’ are to be accounted for. The noun thal ‘mishap, ... ’ is associated
with a Chinese character that has the traditionally assigned reading i, but both the sound and the
meaning are peculiarly Korean, so the word is not to be taken as part of the Chinese vocabulary. The
origin seems to be unknown but it is probably the same as the word thal ‘karma’ attested {"tha'l ol) in
1462 'Nung 8:78b, of unknown etymology. The modern meaning is attested from 1785.
A Reference Grammar of Korean
PARTI 113
4.6. Chinese characters.
To write most words of the Chinese vocabulary the Koreans have traditionally used Chinese
characters, called Hanqca or Hanmunq-ca. For each syllable of a Chinese word there is an
appropriate traditional character, so that knowing the characters is often a help in finding out what
morphemes make up the word. The bulk of the Chinese vocabulary consists of binoms - two-
morpheme (hence two-syllable) words. You can suspect that you are hearing a Chinese word, and
accordingly that the word can be written in Chinese characters, whenever you hear a word that consists
of any two syllables listed in the shape-type table. Sometimes, of course, you may be wrong, especially
if one of the syllables (such as ka or sa) is of a very common type anyway. You would be mistaken to
think that salang ‘love’ should be written with Chinese characters. And sayngkak ‘thinking’, despite
its definitely Chinesey flavor (and perhaps even a Chinese etymology), is not written with characters.
The Chinese characters are listed in dictionaries, and the dictionaries are usually organized
according to a somewhat arbitrary system that analyzes the structure of each character so as to find a
“radical”, traditionally the element that gives the character its category of meaning, and a residual part
that is often called the “phonetic” because it hints at the pronunciation. The 214 Radicals are ordered
according to the number of strokes originally made in writing them (some are now written with fewer
strokes than the order implies), and each of the characters is listed according to its Radical number +
the number of residual strokes. For example, the character may ‘plum’ is listed under Radical 75
(the Tree or Wood Radical), in the subgroup of characters that have a residual-stroke count of 7, so
that we can designate its general location as “75.7”. The radical, as so often, is on the left and by itself
is a character that means ‘tree’. The part on the right is the phonetic; by itself it is the character may
‘every’ which is listed under Radical 80, while the bottom part of that character by itself is mo
‘mother’ {§:• It sometimes happens that a character has a residual-stroke count of zero; that is, the
character is a radical itself, like mok ‘tree’ 75.0. When there are several characters with the same
residual-stroke count, the order is usually determined by an arbitrary tradition that follows earlier
dictionaries. Most Korean dictionaries of Chinese characters also have an index by “total stroke
count”, which is useful if the radical is not readily ascertainable, and an index by Korean readings
arranged according to the Hankul alphabet, with the order under each reading determined either by the
radical or by the total stroke count. A list of the names by which Koreans call the 214 Radicals will be
found in Appendix 5. (Some of the names, like the radicals they represent, are rare.)
When we look up a character in a Chinese character dictionary, called Hanqca sacen or (after the
name of a dictionary famous in ancient China) okphyen “Jewel Book”, we are given only scanty
information, usually just a Korean “reference tag” which tells us the appropriate pronunciation and
something of the meaning. If there are several meanings or readings, each is given, usually following
the traditional entries of large Chinese dictionaries, so that the information is often archaic and not
always relevant to real loanwords that are used in Korean today. The typical form of a reference tag
for a noun is the Korean translation (as a noun or noun phrase) + the reading: A salam in ‘the in that
means salam “man”’. The reference tag for a verb typically gives the Korean translation in the form of
a prospective modifier (or a phrase that ends in one) + the reading: jf, polq kyen ‘the kyen that
means pol “to see”’. The prospective modifier is also used for the adjectives, but a few of the
adjectives are tagged with the simple modifier, to differentiate them from processive use of the same
stem: zK khun tay ‘the tay that means khun “big”’ < khun tay (1576 'Yuhap 2:47b) vs khulq tay
(Kim Sepcey 1957:59) ‘the tay that means “to grow”’; M kin cang ‘the cang that means “long”’ <
["]kin tyang < ttyang (1576 ’Yuhap 2:48a) vs [ Jkilq fjtyang < "ttyang (ibid.) ‘to grow’ (replaced
by cala-). Notice that while often a comma sets off the reading from the gloss (“salam, in”, “pol,
kyen”, “khun, tay”) it is usual to pronounce the tag without a pause, and the prospective modifier
regularly carries the reinforcing -q.
The table on the next two pages is a continuation of §4.5. from the preceding page.
114 PARTI
A Reference Grammar of Korean
[continuation of §4.5]
SHAPES OF CHINESE MORPHEMES
a ya wa e ye - ay way ey - yey o yo
oy
wu
yu
- uy
i wi
ang
yang
wang
yeng - ayng
-
-
*
ong
yong
-
wung
yung
ung -
ing
ak
yak
-
ek yek - ayk
-
-
ok
yok
-
wuk
yuk
-
ik
am
-
-
em yem
-
-
-
-
-
-
-
-
um
im
ap
-
-
ep yep
-
-
-
-
-
-
-
-
up -
ip
an
-
wan
en yen wen -
-
on
-
-
wun
yun
un
in
al
•
wal
el yel wel -
“
■
”
ol
•
*
wul
yul
ul
il
ka
.
kwa
ke - - kay
kwa
key
kwey kyey
ko
kyo
koy
kwu
kyu
.
ki kwi
kang
-
kwang
kyeng - -
-
kong
koyng kwung
-
kung
-
kak
kyak
kwak
kyek - -
-
kok
koyk
kwuk
-
kuk
(k)kik -
kam
•
-
kern kyem - -
•
-
-
-
-
kum
(kirn) -
kap
-
-
kep kyep - -
-
-
-
-
-
kup
-
kan
-
kwan
ken kyen kwen -
-
kon
-
kwun
kyun
kun
kin
kal
-
kwal
kel kyel kwel -
-
kol
-
kwul
kyul
-
kil
*
-
-
- •
khway
-
-
-
-
■
-
-
ha
hwa
he - - hay
hway
hwey hyey
ho
hyo
hoy
hwu
hyu
-
huy -
hwi
hang hyang hwang
hyeng - hayng
-
-
hong
-
hoyng hwung
hyung hung
- -
hak
hwak
hyek - hayk
-
-
hok
-
hoyk
-
hyuk
huk
- -
ham
-
hem hyem - -
-
-
-
-
-
-
-
hum
- -
hap
-
- hyep - -
-
-
(hop)
-
-
-
-
hup
• -
han
hwan
hen hyen hwen -
-
-
hon
-
-
hwun
-
hun
- -
hal
hwal
hel hyel - -
-
-
hoi
-
-
hwul
hyul
hul
- hil
ma
-
*
- may
myey
mo myo
mu
-
•
- mi
mang -
-
myeng
- mayng -
•
mong -
-
-
•
-
mak
“
- myek
- mayk
: : :
mok
muk
:
:
: :
man
.
- myen
. .
...
. . .
mun
.
.
- min
mal
-
- myel
* *
-
mol
mul
-
-
- mi l
*
- pay
*
"
po
*
*
pu
*
p>
pang
-
pyeng
“ *
*
-
pong
-
pung
-
ping
pak
- pyek
pern
- payk
.
-
pok
:
puk
:
:
pan
.
pep -
pen pyen
.
.
.
pon
.
pun
.
pin
-
pal
“
pel pyel
* *
'
'
*
*
pul
*
pha
_
. „
- phay
.
.
phyey pho
phyo
.
phi
_
*
*
-
- phyeng
- phayng
-
-
-
-
*
phung -
“
-
-
phyak
-
-
- phayk
-
*
phok
-
phuk
-
-
-
■
-
- phyem
' -
-
-
-
phum
-
-
-
*
-
“ -
-
-
-
-
-
phip
phan
-
*
phyen
- -
-
-
-
(phun) -
*
-
phal
-
*
- *
*
*
*
phi l
-
-
-
A Reference Grammar of Korean
PARTI 115
na
•
mye
-
nay
-
no
nyo
noy
nwu
-
-
ni
nang
-
nyeng
-
-
-
nong
-
-
-
nung
-
nak
-
?nyek
-
-
-
•
-
-
nyuk
nik
nam
-
nyem
-
-
-
-
-
-
-
nim
nap
-
?nyep
-
-
-
-
-
-
-
-
nan
-
nyen
-
-
-
-
-
nwun
-
-
nal
?nwal
nyel
*
*
•
nwul
•
ni l
la
.
.
lye
_
lay
.
lyey
lo
lyo
loy
Iwu
lyu
.
li
lang
lyang
-
lyeng
-
layng
-
long
lyong
-
lyung
lung
-
lak
lyak
-
lyek
-
-
•
lok
-
lyuk
luk
-
lam
-
-
lyem
-
-
-
-
-
-
lum
lim
lap
•
-
lyep
-
-
-
-
-
-
-
lip
lan
-
-
lyen
-
-
-
Ion
-
lyun
-
lin
lal
'
lyel
*
“
*
lyul
■
ta
.
.
.
.
tay
.
.
to
.
toy
twu
.
.
tang
-
tek
-
-
-
-
tong
-
-
-
tung
-
-
-
-
-
-
tok
-
-
-
tuk
tarn
-
-
-
-
-
-
-
-
-
-
tap
-
-
-
-
-
-
-
-
-
-
tam
-
-
-
-
-
ton
-
twun
-
-
tal
•
■
*
•
*
tol
*
•
■
■
tha
_
the
thay
_
tho
thoy
thwu
_
thang
-
thayng
-
thong
-
-
-
-
-
-
thak
-
thayk
-
-
-
-
thuk
-
-
tham
-
-
-
-
-
-
thum
-
-
thap
-
-
-
-
-
-
-
-
-
than
-
-
-
thon
-
-
-
-
-
thal
-
-
-
-
-
-
-
-
-
-
-
-
ca
cwa
ce
- cay
cey -
-
CO
coy
cwu
-
- Cl
-
cang -
ceng -
- cayng -
-
-
cong -
cwung -
cung
- cing
-
cak -
cek
-
. *p
cok -
cwuk -
cuk
- cik
-
cam
cem
•
-
-
-
?cum
- cim
-
cap
cep
-
-
-
-
cup
“ cip
-
can
cen
•
-
con
cwun -
-
- cin
-
-
cel
-
-
col
?cwul -
-
-
-
cha
-
che
- chay chway chey -
-
cho
choy
chwu -
chi chwi
chang -
-
cheng -
- chayng - -
-
chong -
-
chwung -
chung -
ching -
chak -
-
chek -
. . - -
-
chok -
-
chwuk -
chuk -
ch i k
cham -
-
chem -
- -
-
-
-
-
-
chim
-
-
chep -
- - . -
-
-
-
-
-
chip
chan -
-
chen -
- -
-
chon -
-
chwun -
(chun) -
chin
chal -
chwal
chel -
- -
-
-
-
chwul -
-
ch i l
sa
se
- say
sway
sey -
so
soy
swu
-
- SI,SSI
sang,ssang -
seng -
- sayng
-
-
song -
swung -
sung
-
sak
sek -
- sayk
-
-
sok
swuk
-
- sik
sam
sem
-
-
-
-
-
-
- sim
sap
sep
-
-
-
-
-
sup
- sip
san
sen
-
-
-
son
swun
-
- sin
sal
sel
-
-
-
sol
swul
sul
- si l
116 PART I
A Reference Grammar of Korean
4.7. Characters with multiple readings.
For each Chinese character there is usually one basic “reading” (= pronunciation). This means
that each character represents one loanmorph in the Chinese vocabulary; of course the pronunciation is
subject to the same sort of automatic alternations as any other element in Korean (-p becomes -m
before m-, etc.). But some characters have two, and rarely three, readings that cannot be predicted
except by knowing the particular words in which they are used. There are two types of multiple
readings. In the one type, difference of meaning goes with the difference in shape, and we have two
loanmorphs represented by a single Chinese character with usually a single etymological origin. In the
other type, there is no difference in meaning; we have variant versions of the same loanmorph. There
are a few cases that are simply variants that have been spawned for the whole word, and those we have
not listed below: phyengphung for pyengphung assimilated the first syllable to the aspiration of the
second; ‘nachim-phan for ‘nachim-pan ‘compass’, where the aspiration of the last syllable comes
from assimilating to the preceding syllable, is the standard form in the North Korean dictionary. Nor
have we included khan, which seems to be a nonstandard variant of kan ‘interval’.
In the list of the first type of multiple readings, the reference tag is shown for each reading and
critical examples are given. In the second list, the meaning of the character is given on the left; then
critical examples are listed and the relevant morph shapes are presented in small capital letters, with
the more common shape given first. There follows a third list of a few morphemes with unpredictably
varying shapes that begin with basic 1™ or n- (Cf §4.4, §4.8).
4.7.1. Multiple readings: list one.
CHARACTER
SHAPES
TAGS
EXAMPLES
1.3 fl
chwuk/chwu
so chwuk ‘sign of Ox’
chwuk-si ‘Hour of the Ox’
3.3 ft
ilum chwu (in names)
Kong Songchwu (person)
tan / lan
pulk.ulq tan ‘red’
tanswun ‘red lips’
tansim ‘sincerity’
molan (uy) lan ‘peony’
molan ‘peony’
5.10f2
ken / kan
hanul ken ‘Heaven (as
ken-kon ‘Heaven and Earth’
9.7 if
divination symbol)’
malulq kan ‘be dry’
kanco = kenco ‘drying’ 1
phyen /pyen
phyen-hal phyen
phyen.li ‘comfort’
‘comfortable’
taysopyen (uy) pyen
taypyen ‘defecating’
18.7 10
‘easing nature’
sopyen ‘urinating’
cuk/cik
kot cuk ‘id est’
yencuk ‘if so’
19.2 ft
pepchik chik ‘rule’
pepchik ‘rule’
kyuchik ‘regulation’
pun / phun
nanwulq pun ‘divide;
punswu ‘fraction’
minute’
pun swu ‘no. of minutes’
21.3 it
ton phun ‘farthing’
il-phun ‘one farthing’
puk/pay
puk-nyekh puk ‘north’
nam-puk ‘North and South’
phay-halq pay ‘suffer
defeat’
phaypay ‘defeat’
28.9 #
cham/sam
chamka hal cham
chamka ‘participation’
‘participate’
pyelq ilum sam
samseng ‘21st of the 28
(constellation)
Constellations’
30.3
hap/hop
hap-hal hap ‘join’
haptong ‘combination’
hop hop (measure)
il-hop ‘one hop (a
third of a pint)’
A Reference Grammar of Korean
PARTI 117
30.4
pO/pi
30.6 P0
in / yel
32.101H
sayk/say
37.6 M
kyey / ke
38.8
pha / pa
40.8 It
swuk / swu
50.2 ft
pho / po
53.6 m
to / thak
53.11 )fj?
kwak/hwak
60.6 ^
sol/lyul
See §4.8 (p.
60.9 U
pu / pok
60.12jfj
dng / chi
61.8 jg
ak/o
ani pO ‘not’
mak.ulq pi ‘clog up’
mok-kwumeng in ‘throat
mok-meyilq yel ‘choke’
mak.ulq sayk ‘block,
stop up’
pyenpang say ‘fort’
kyeyyak kyey
‘contract’
Kelan (uy) ke
(phonetic)
nulk.un kyeycip pha
Sapa-seykyey pa
cam-calq swuk
‘stay overnight’
pyel swu
‘constellation’
pellil pho ‘spread’
ton po ‘alms’
pep to ‘law; degree’
heyalil thak
‘estimate’
tey-twuli kwak
‘enclosure’
pil hwak ‘empty’
kenulilq sol ‘command’
pl'yul *yul ‘ratio’
tasi pu ‘again’
hoypok-halq pok
‘recover’
puluq ring ‘recruit’
um.ak chi ‘4th note
of pentatonic scale’
modi ak ‘bad’
miwe hal o ‘hate’
pOkyel ‘voting down’
pisayk ‘frustrated by
fortune’
’ inhwu ‘throat’
o.yel ‘sobbing’
censayk ‘obstruction’
cilqsayk ‘disgust’
saykchayk ‘sidestepping
responsibility’
sengsay ‘fortress’
yosay ‘fortress’
kyeyyak ‘contract’
kyeyki ha- ‘reach deadline’
Kelan ‘Khitan Tatars’
'nopha ‘old woman’
Sapa ‘Saha; This World’
swukpak ‘lodging’
sengswu ‘the stars’
Isip-phalq swu ‘the 28
Constellations’
phoko ‘decree, proclamation’
posi ‘Buddhist almsgiving’
"yento ‘year period’
cengto ‘degree, extent’
Vothak ‘conjecture’
yothak ‘mental telepathy’
chonthak ha- ‘surmise’
sengkwak ‘castle walls’
hwakcheng ‘purification’
thongsol ‘general command’
insol ‘leading (people)’
kwensol ‘the family one
heads’
pi'yul ‘ratio’
nung.lyul ‘efficiency’
puhwal ‘resurrection’
puhung ‘revival’
wangpok ‘round trip’
pok.kwu ‘restoration’
sangcing ‘symbol’
cingpyeng ‘conscription’
sen-ak ‘good and/or bad’
ak.han ‘villain’
cungo ‘hatred’
ho-o ‘likes and dislikes’
118 PART I
A Reference Grammar of Korean
64.8
chwu / thoy
kalul chwu ‘discriminate’
ssah.ul thoy ‘accumulate’
64.9 £§
cey / li
pachilq cey ‘offer’
Poli (uy) li (phonetic)
66.11
swu / sak / chok
seym swu ‘number’
cacwu sak ‘frequent’
ppaykppayk hal chok
‘dense’
72.4 Jg
yek/i
pakkwulq yek ‘change’
swiwulq i ‘easy’
73.3 I!
kayng/kyeng
tasi kayng ‘again’
kochilq kyeng
‘change’
75.11 m
lak / ak
culkil lak ‘rejoice’
um.ak (uy) ak ‘music’
79.6 $
sal/sway
cwuk.ilq sal ‘kill’
sangsway (uy) sway
‘attack’
85.4 ?£
chim/sim
camkil chim ‘sink’
ilum sim (name)
85.5
pi / pul
kkulh.ulq pi ‘boil’
saym sos.ulq pul
‘jet’
85.10 ?f
hwal / kol
mikkulewul hwal
‘slippery’
iksal kol ‘humor’
94.4 jK
sang/cang
mo.yang sang
‘appearance’
kulq-cang cang
‘document’
94.7 Jfc
hyep/hap
cop.ul hyep
‘narrow’
ilum hap (in names)
102.7 H
hwa/hoyk
kulim hwa ‘drawing’
kul-ssi hoyk
‘brush stroke’
106.0 &
payk/pay
huyn payk ‘white’
Paychen pay (name) 2
chwutan ‘judgment’
thoycek ‘accumulation’
ceychwul ‘presentation’
Poli ‘Bodhi, Buddhahood’
swuqca ‘numeral’
™(-q) swu ‘number of ’
pinsak ha- ‘be frequent’
saksak ‘constantly’
sakchey ‘constant shifting
of personnel’
chok.ko ‘fine mesh’
mu.yek ‘trade’
Yekse ‘the Book of Changes’
yongi ha- ‘be easy’
kayngsin ‘renovation’
kayngsayng ‘rebirth’
pyenkyeng ‘change’
kyengcil ‘change (in
structure)’
'nak.wen ‘paradise’
um.ak ‘music’
sal.in ‘murder’
sangsway ‘counter¬
balancing’
swayto ‘onslaught’
kamsway ha- ‘impair’
chimmol ‘sinking’
Sim ssi ‘Mr Shim’
pitung ‘boiling’
pul.yen ha- ‘be quick¬
tempered’
hwalqsek ‘talcum’
kolkyey ‘humor’
sangthay ‘state, condition’
hyengsang ‘form’
sang(q-)cang ‘citation
of merit’
kongkayq-cang ‘open
(public) letter’
hyepchak ‘narrowness’
Hapchen (place)
hwaka ‘artist’
hoyk swu ‘stroke count’
paykpal ‘white hair’
Paychen onchen ‘Paychen
hot springs (spa)’
A Reference Grammar of Korean
PART I 119
109.4 %
seng/sayng
salphilq seng
sengchal ‘reflection’
‘investigate’
sengmyo ‘visiting
ancestral tombs’
kwuk.mu seng ‘State
Department’
Santong seng ‘Shantung
(Shandong) Province’
120.4 ^
telq sayng ‘lessen’
kamsayng ‘curtailment’
sayng.lyak ‘abbreviation’
sayk/sak
chac.ulq sayk ‘seek’
swusayk ‘search’
mosayk ‘groping’
sasayk ‘speculation’
sayk.in ‘index’
ssulssul-halq sak
sak.yen hata ‘is lonesome’
‘lonesome’
cwul sak /sayk
chelqsa(y)k ‘cable’
140.9 ^
ce/chak
ciulq ce ‘create’
cese ‘written works’
ceca ‘author’
putic.chil chak ‘hit’
chak.lyuk ‘landing’
chakswu ‘putting one’s
hand to, beginning’
140.9 M
yep/sep
iph(-sakwi) yep ‘leaf’
ci-yep ‘branches and
leaves; minor details’
Hum sep (in names)
Sep ssi ‘Celcius’
Kasep(-wen) ‘Kasapa
(plain)’
144.0 'a
hayng/hang
tanil hayng ‘go; do;
Vehayng ‘travel, a trip’
market’
unhayng ‘bank’
- hayng ‘bound for ’
hayng.lyel 3 ‘procession’
hang.lyel hang ‘degree
hang.lyel 3 ‘degree
145.4 n
of relationship’
of relationship’
soy/choy
yak-halq soy ‘weak’
soyyak ‘debilitation’
sangpok choy ‘mourning
chamchoy (a kind of
147.0 J|
garb’
mourning garb)
kyen/hyen
polq kyen ‘see’
kyenhay ‘opinion, view’
nathanal hyen ‘appear’
alhyen ‘royal audience’
149.7
sel/sey/yel
mal-halq sel ‘speak’
selhwa ‘narration; sermon’
tallaylq sey ‘coax’
yusey ‘electioneering’
ilumq yel (in names)
Kim Sam.yel (person)
149.15
tok/twu
ilk.ulq tok ‘read’
tokse ‘reading (books)’
kwicel(q) twu ‘phrase’
kwutwu ‘punctuation’
l Itwu ‘Idu’ 4
157.5 m
pha / phi
celttwuk-pal.i pha
phahayng ‘limping’
‘lame’
kiwul.ye sul phi
phllip ‘standing on one leg’
‘lean to one side’
167.0 ^
kum / kim
hwangkum kum ‘gold’
(hwang)kum ‘gold’
ilum kim (in names)
Kim ssi ‘Mr Kim’
Kimhay (city)
120 PART I
A Reference Grammar of Korean
170.6 kang/hang
213.0 Hij kwi/kyun/kwu
naylilq kang ‘descend’
hangpok hang ‘surrender’
kepuk kwi ‘tortoise’
son the-cilq kyun
‘chapped’
ilum kwu (in names)
kangha ‘descent’
sungkang-ki ‘elevator’
hangpok ‘surrender’
kwisen ‘tortoise-shaped
boat’
kyun.'yel ‘fissure’
*Yi Cengkwu (person)
Kwupho (place)
1 Popularly ken in the meaning ‘dry’ too, as in ken-cenci ‘dry battery’ (Ceng Insung 225).
2 See also paynge < pay-nge ‘whitebait’ (§3.6).
3 Identical morpheme lyel; minimal contrast of hayng and hang.
4 Chinese characters once used to write Korean particles, endings, and the like.
4.7.2. Multiple readings: list two.
CHARACTER
SHAPES
MEANING
EXAMPLE 1
EXAMPLE 2
1.4 ^
pul/pu 1
not, un-
pulphyen
‘discomfort’
pul-kongphyen
‘unfair’
putang ‘injustice’
puceng ‘uncertainty’
9.2 if
cup/cip
ten, some
cupki ‘furniture’
cip.mul ‘furniture’
9.5 ft
cwa / ca 2
assist
pocwa ‘assistance
to superior’
capan ‘salted fish
or caviar’
11.2 ft
nay / na
inside
nayoy ‘in and out’
nain ‘court lady’
12.0 A
phal / pha 3
eight
phal-wel ‘August’
(sa-wel) pha-il
‘Buddha’s birthday’
12.2 7\
'yuk/'yu/nyu
six
Vuk.il ‘6th day’
6lyuk.il ‘5th or
6th day’
'yu’-wel ‘June’
onyu’-wel ‘May or
June’
18.2 $)
cel / chey
cut
celqtan ‘amputation’
kancel ‘eagerness’
ilchey ‘altogether
(< one cut)’
18.6 M
ca / chek
stab
cakuk ‘stimulation’
cakayk ‘assassin’
cheksal ‘stabbing
to death’
24.0 +
sip/si’
ten
sip-il ‘eleven’
si’-wel ‘October’
29.2 £
pan / pen
reversal
pantay ‘opposition’
pentap ‘turning it
into rice land’
pencen ‘converting
rice land back’
30.2 ft)
kwu / kwi
sentence
kwutwuq-cem
‘punctuation mark’
kwicel ‘couplet’
40.3 %
thayk/tayk
house
kathayk ‘domicile’
cwuthayk ‘residence’
tayk ‘your house’
si-tayk ‘husband’s
house’
48.7 ^
cha / chi
difference
chai ‘difference’
chamchi(-pucey) ‘lack
of uniformity’
50.5 te
chep / chey
document
swuchep ‘notebook,
album’
cheyci, (cheymun)
‘document of
appointment’
64.8 W-
chel / chey
restrain
chelqcwu ‘hindrance’
?cheyli 4 ‘restraint’
A Reference Grammar of Korean
PART I 121
68.9 m
ci m / chi m
guess
cimcak ‘conjecture’
chim.lyang (same)
72.11 H
phok/pho
violent
phoktong ‘riot’
phokphung ‘tempest’
phokto ‘rioters’
hoyngpho ‘tyranny’
phohak ‘tyranny’
(photo ‘rioters’)
phoki = capho-(caki)
‘despair’
72.15Bi
phok / pho
expose
phok.yang ‘burning
sun’
phopayk ‘bleaching
in the sun’
75.0 Tfc
mok / mo
tree, wood
mok.kun ‘tree root’
mokwa ‘papaya’
(< mok-kwa)
85.6 M
tong/thong
alley;
clear
tongkwu ‘village’
thongchal, thongchok
‘discernment’
107.0 BL
phi / pi
skin
phipu ‘skin’
'nokpi ‘deerskin’
115.5
phyeng/ching
balance
chenphyeng ‘balance,
scales’
chingchwu ‘balance
weight’
119.10 if
tang / thang
sugar
tangpun ‘sugar
content’
photo-tang ‘glucose’
selthang
‘granulated sugar’
sathang ‘sugar;
sweets’
140.6 ft
ta / cha
tea
tapang ‘teahouse’
takalq-sayk ‘light
brown’
(hong)cha ‘(black)
tea’
(chaq-pang, chaq
cip) -» tapang
‘tearoom’
149.7 ^
se/sey
swear
se.yak ‘oath’
mayngsey ‘pledge’
149.1211
sik/ci
knowledge
cisik ‘knowledge’
phyoci = phyosik
‘mark, signal’
159.0 $
cha / ke
vehicle
cha ‘vehicle, car’
kicha ‘train’
catong-cha ‘auto’
cacen-ke ‘bicycle’
cengke-cang ‘rail
station’
161.0 Jg
cin / sin
Sign of
the Dragon
cin-si ‘Hour of the
Dragon’
thansin ‘birthday’
184.0 &
sik/sa
eat
umsik ‘food’
tansa ‘lunch-basket
rice’
187.4 K
thay/tha
stupid;
burden; ...
thaycak ‘worthless
work’
si-tha ‘load’ 1 2 3 4 5
1 The ™1 drops regularly before t or c (Cf §3.5).
2 CF §2.7.4.
3 Cf §3.5.
4 1 Yi Ungpayk 571. I cannot find the word in any dictionary, nor can I find ?cheycen ‘lightning-
fast’, a purported example that comes from an unknown source.
5 But usually pronounced si-thay. In fact, does anybody say /sitha/? I assume (perhaps wrongly) an
etymology that involves both sH-/sil- ‘load’ and the Chinese loanmorph: si[t]-tha(y). The final -
y is etymological and not a reduction of the incorporated i that is common for nouns in certain
northern dialects. The si-tha version is not found in the major dictionaries; is it a ghost? Kim
Minswu and Hong Wungsen treat thaycak as nonstandard for thacak, a version that I have not
found in the other dictionaries.
122 PART I
A Reference Grammar of Korean
4.7.3. Multiple readings: list three.
CHARACTER
SHAPES
MEANING
40.11
nyeng / lyeng
< NYENG
calm
61.5 &X
no/lo
< "NWO
anger
149.8 f&
Ion / non /('non)
< LWON
discussion
149.9 fg
?nak / lak
< 'NAK
acquiesce
EXAMPLE 1
EXAMPLE 2
annyeng ha- 1 ‘be in
milyeng ‘illness,
good health’
cengnyeng ‘for sure’
indisposition’
no ha- ‘get angry’
pGnno ‘indignation’
kyek.no ‘wild rage’ 2
taylo ‘great anger’
'lion ‘logic’
uynon ‘discussion’ 3
lion ‘dissent’
'nonmun ‘treatise,
en.lon ‘discussion’
mullon 4 ‘of course’
dissertation’
sungnak 5 ‘consent’
helak ‘permission’
khwaylak ‘ready
assent’
1 But those speakers who say /allyeng/ are treating this as a single-reading character lyeng. For
them cengnyeng is properly to be analyzed as ceng.lyeng.
2 From this form alone we would not know whether the appropriate spelling of the second element
in /kyengno/ is lo or no. The practice is to write lo only when /l/ is pronounced. Notice that
/punno/ could not be “pun.lo” for that would be pronounced /pullo/.
3 *Yi Ungpayk draws a distinction between uynon ‘discussion’ and uylon ‘argument’ and on that
both NKd and LHS agree, but KEd puts the two words together as uy'non and gives uylon as a
variant pronunciation of uynon, a shape that is unanticipated, in any event, though attested
from at least the late 19th century: uynwon ho.ye la (1893 Scott 4) ‘consult!’.
4 This could, of course, be from “mul.'non but we decide on Ion for reasons of history and the
relative infrequency of 'non in environments where it would be distinguishable.
5 I see no reason we cannot write sung.lak and say the character has only the reading lak with
/nak/ as an automatic alternant. If some speakers say (?)khwaynak, however, that is another
matter.
A Reference Grammar of Korean part i 123
4.7.4. Index to the lists of multiple readings.
ak / lak
75.11
II
hyen / kyen
147.0
I
ak/o
62.8
II
hyep/hap
94.7
I
ca / chek
18.6
II
i/yek
72.4
I
ca/cwa
9.5
I
in/yel
30.6
I
cang / sang
94.4
I
kan / ken
5.10
I
ce/chak
140.9
II
kang/hang
170.6
I
cel / chey
18.2
I
kayng/kyeng
73.3
I
cey / li
64.9
II
ke/cha
159.0
II
cha / chi
48.7
II
ke/kyey
37.6
I
cha / ke
159.0
II
ken / kan
5.10
I
cha / ta
140.6
I
kim/kum
167.0
I
chak / ce
140.9
I
kol / hwal
85.10
I
cham / sam
28.9
II
kum / kim
167.0
I
chek/ca
18.6
II
kwak / hwak
53.11
I
chel/chey
64.8
II
kwi / kwu
30.2
II
chep / chey
50.5
II
kwi / kyun / kwu
213.0
I
chey / cel
18.2
II
kwu / kwi
30.2
II
chey / chel
64.8
II
kwu/kwi/kyun
213.0
I
chey/chep
50.5
II
kyen /hyen
147.0
I
chi/cha
48.7
I
kyeng/kayng
73.3
I
chi / ting
60.12
I
kyey/ ke
37.6
I
chik / cuk
18.7
II
kyun / kwi / kwu
213.0
I
chim/cim
68.9
I
lak/ak
75.11
I
chim / sim
85.4
II
lak/?nak
149.9
III
ching / phyeng
115.5
I
lan / tan
3.3
I
chok/swG/sak
66.11
I
li / cey
64.9
I
choy / soy
145.4
I
lo/no
61.5
III
chwu / chwuk
1.3
I
Ion / 'non
149.8
III
chwu / thoy
64.8
I
lyeng / nyeng
40.11
III
chwuk / chwu
1.3
II
mo / mok
75.0
II
ci / sik
149.12
II
mok / mo
75.0
II
tim/chim
68.9
II
na / nay
11.2
II
tin /sin
161.0
I
?nak / lak
149.9
III
ting / chi
60.12
II
nay / na
11.2
II
tip /cup
9.2
I
no/lo
61.5
III
cuk / chik
18.7
II
'non / Ion
149.8
III
cup / tip
9.2
I
nyeng / lyeng
40.11
III
cwa/ca
9.5
II
nyu/'yuk/'yu
12.2
II
hang / hayng
144.0
I
o/ak
61.8
I
hang / kang
170.6
I
pa/pha
38.8
I
hap/hop
30.3
I
pan / pen
29.2
II
hap/hyep
94.7
I
pay/payk
106.0
I
hayng / hang
144.0
I
pay/puk
21.3
I
hop/hap
30.3
I
payk / pay
106.0
I
hoyk / hwa
102.7
I
pen / pan
29.2
II
hwa / hoyk
102.7
I
pha / pa
38.8
I
hwak / kwak
53.11
I
pha / phal
12.0
II
hwal / kol
85.10
I
pha / phi
157.5
I
124 PART I
A Reference Grammar of Korean
phal / pha
12.0
II
sep/yep
140.9
I
phi / pha
157.5
I
sey/se
149.7
II
phi / pi
107.0
II
sey/sel/yel
149.7
I
pho/phok
72.11
II
si / sip
24.0
II
pho/phok
72.15
II
sik/ci
149.12
II
pho / po
50.2
I
sik/sa
184.0
II
phok/pho
72.11
II
sim/chim
85.4
I
phok/pho
72.15
II
sin/cin
161.0
II
phun/pun
19.2
I
sip/si
24.0
II
phyen/pyen
9.7
I
sol/'yul
60.6
I
phyeng/ching
115.5
II
soy/choy
145.4
I
pi / phi
107.0
II
sway / sal
79.6
I
pl/pu
30.4
I
swu/sak/chok
66.11
I
pi / pul
85.5
I
swu / swuk
40.8
I
po / pho
50.2
I
swuk/swu
40.8
I
pok / pu
60.9
I
ta / cha
140.6
II
pu/pl
30.4
I
tan / lan
3.3
I
pu/pok
60.9
I
tang / thang
119.10
II
pu/pul
1.4
II
tayk/thayk
40.3
II
pul/pi
85.5
I
tha/thay
187.4
II
pul / pu
1.4
II
thak/to
53.6
I
pun / phun
19.2
I
thang / tang
119.10
II
puk/pay
21.3
I
thay/tha
187.4
II
pyen /phyen
9.7
I
thayk/tayk
40.3
II
sa/sik
184.0
II
thong/tong
85.6
II
sak/sayk
120.4
I
thoy / chwu
64.8
I
sak/swu/chok
66.11
I
to / thak
53.6
I
sal / sway
79.6
I
tok/twu
149.15
I
sam / cham
28.9
I
tong / thong
85.6
II
sang/cang
94.4
I
twu / tok
149.15
I
say/sayk
32.10
I
yek/i
72.4
I
sayk/sak
120.4
I
yel /in
30.6
I
sayk/say
32.10
I
yel/sel/sey
149.7
I
sayng/seng
109.4
I
yep/sep
140.9
I
se/sey
149.7
II
'yu/'yuk/nyu
12.2
II
sel/sey / yel
149.7
I
'yuk/'yu/nyu
12.2
II
seng/sayng
109.4
I
'yul/sol
60.6
I
4.8. Chinese morphemes with basic 1-.
A number of problems arise with morphemes which had a basic 1- initial in Middle Chinese. In
South Korea the standard practice is to spell these morphemes in two ways: with n- (or zero before i,
y) if they are at the beginning of a word, with 1- elsewhere. In North Korea the morphemes are
spelled with initial 1— in all positions. Our Romanization writes 'n- (or l - before i, y) for these cases
where the two systems diverge. There are a few morphemes which, despite an etymological !•••, are
standardized to n~ in both Koreas: no = 'no ‘oar’, nwu = 'nwu ‘loft, pavilion’, nwu = 'nwu
‘frequent’ (Cf nwunwui = 'nwu'nwu-i ‘frequently’), nwuki = 'nwuki ‘dampness’. In the word
sil'yen ‘disappointment in love’, we find an unusual shortening of the expected sillyen, ignored in the
North Korean spelling, but indicated in South Korea by writing sil.yen. A similar case should be
no'yey /no(.y)ey/ ‘slave’ (Cf halyey ‘male slave’, kwan.lyey ‘official slave’, 'yeysok ‘subordination’),
but the North Korean spelling seems to be no.yey, like that of the south.
A Reference Grammar of Korean
PART I 125
A small group of morphemes have the shapes lyul and lyel. These morphemes follow a special
pronunciation rule: after a vowel or n, the 1 unexpectedly drops. The liquid actually appears only after
1 or some consonant other than n (when it appears in the reflex Ini), so that we would not know that
these morphemes begin with a basic 1 (rather than n) without additional information from their history
or from dialect pronunciations. The alternation is ignored in North Korean spelling, but noted by the
South Koreans, so we mark it by —*y •• in our Romanization. Examples:
CHARACTER
AFTER VOWEL
AFTER -n
AFTER -1
ELSEWHERE
LYUL
60.6#
kyu'yul
sen.'yul
illyul
pep.lyul
< ' LYWULQ
rule
/kyuyul/
‘discipline’
/senyul/
‘rhythm’
/illyul/
‘uniformity’
/pemnyul/
‘law’
95.6 $
plVu 1
hwan.'yul
kwulqcel-lyul
nung.lyul
ratio
/plyul/
‘ratio’
¥yul
/lyul/
‘interest rate’
/hwanyul/
‘exchange rate’
/kwulccellyul/
‘index of
refraction’
‘efficiency’
75.6 H
phi'yul
san^yul
hallyulq-sek
hwang.lyul
chestnut
/phiyul/
‘unshelled
chestnut’
/sanyul/
‘Japanese
chestnut’
/hallyulssek/
‘stones cut to
chestnut size’
/hwangnyul/
‘dried peeled
chestnuts’
LYEL
19.4
wu'yel
chen^y el
collyel
yong.lyel
< LYELQ
inferior
/wuyel/
‘superiority and
inferiority’
pfyel
/piyel/
‘baseness’
/chenyel/
‘lowly, humble’
/collyel/
‘clumsiness’
/yongnyel/
‘inferiority’
i8.4 y\\
'na.Vel
pan^yel
illyel
hayng.lyel
rank, order
/na.yel/
‘array’
/panyel/
class rank’
/illyel/
‘a line’
/hayngnyel/
‘procession’
86.6
uy'yel
sen.'yel
yellyel ha-
mayng.lyel
fierce
/u.yel/
‘heroism’
/senyel/
‘veteran patriot’
/yellyel(h)a-/
‘be ardent’
/mayngnyel/
‘fury’
145.6 gi
pha.Vel
kyun.Vel
kyellyel
cak.lyel
rip
/pha.yel/
‘explosion’
/kyunyel/
‘fissure’
pun^yel
/punyel/
‘disruption’
/kyellyel/
‘rupture’
/cangnyel/
‘explosion’
126 PART I
A Reference Grammar of Korean
4.9. Tongkwuk readings.
vol 1
vol 2
k
kh
kk
ng
t
th
rt
rt
n
P Ph
PP
m
c
ch
cc
s
ss
q
h
hh
_
1
z
1
-ung
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
-ung
+
+
-ung
+
+
+
+
+
+
+
+
+
+
+
+
-uk
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
-ing
+
+
+
+ +
+
+
+
+
+
+
'-ing
+
+
+
+
-ing
+
+
+
+
+
+
+
-ik
+
+
+
+
+
+
+
+
+
-oyng
+
+
+
+
+
+
+ +
+
+
+
+
+
+
+
+
+
-oyng
+
+
+
+
+
+
-oyng
+
+
+
+
+
+
+
+
+
-oyk
+
+
+
+
+
+
+ +
+
+
+
+
+
+
+
+
-woyng
+
+
+
+
+
'-woyng
+
+
-woyng
+
-woyk
+
+
+
+
-wuyng
+
+
-wuyk
+
-wong
+
+
+
+
+
+
+
+
+ +
+
+
+
+
+
+
+
+
+
+
’-wong
+
+
+
+
+
+
+ +
+
+
+
+
+
+
+
+
-wong
+
+
+
+
+
+
+ +
+
+
+
+
+
+
+
+
+
-wok
+
+
+
+
+
+
+ +
+
+
+
+
+
+
+
+
+
-ywong
+
+
+
+
+
+
+
+
+
'-ywong
+
+
+
+
+
+
+
+
+
-ywong
+
+
+
+
+
-ywok
+
+
+
+
+
+
+
+
-ang
+
+
+
+
+
+
+
+
+ +
+
+
+
+
+
+
+
+
+
+
-ang
+
+
+
+
+
+
+
+
+ +
+
+
+
+
+
+
+
+
+
-ang
+
+
+
+
+
+
+
+
+ +
+
+
+
+
+
+
+
+
+
+
-ak
+
+
+
+
+
+
+
+ +
+
+
+
+
+
+
+
+
+
+
-yang
+
+
+
+
+
+
+
+
+
+
+
+
+
+
-yang
+
+
+
+
+
+
+
+
+
+
+
+
-yang
+
+
+
+
+
+
+
+
+
+
+
+
+
-yak
+
+
+
+
+
+
+
+
+
+
+
+
+
+
-wang
+
+
+
+
+
+
+
+
+
+
+
+ +
+
‘-wang
+
+
+
+
+
+
-wang
+
+
+
+
+
+
+
+
+
+
-wak
+
+
+
+
+
+
+
+
+
+
+
+
+
2
-wung
+
+
+
+
wung
-wung
+
-wuk
+
+
+
+
-ywung
+
+
+
+
+
+
+
+
+
+
+
'-ywung
+
+
+
-ywung
+
+
+
-ywuk
+
+
+
+
+
+
+
+
+
+
+
+
k
kh
kk
ng
t
th
tt
n
P Ph
PP
m
c
ch
cc
s
ss
?
h
hh
-
1
z
A Reference Grammar of Korean
PART I 127
(vol 2)
7. -yeng
“-yeng
'-yeng
’ -yek
-yweng
“-yweng
*-yweng
’-ywek
8. -on
“-on
' —on
-un
“-un
* -un
* -ulq
-in
“-in
* -in
’ -Hq
9. -won
“-won
-won
-wolq
-an
-an
-an
-alq
-wan
-wan
-wan
-walq
10
k kh kk ng t
+ + + + +
+ + +
+ + + + +
+ + + + +
+ + + +
+ + +
+
th tt n p ph pp m
+ + + + + + +
+ + + + + + +
+ + + + + + +
+ + + + + + +
c ch cc s
+ + + +
+ + + +
+ + + +
+ + + +
+
ss
+
+
+
+ +
+ + +
+ +
+
+ +
+ +
+ + +
+ + +
+
+ +
+ +
+ +
+ +
+ +
+ +
+ +
+ +
+ +
+ +
+
+ +
+ +
+
+
+
+
+ + +
+ +
+ + +
+ + +
+ + +
+ +
+ + + +
+ + +
+ + + +
+ + + +
+ + + +
+ + + +
+ + + +
+ + + +
+ + + +
+ + + +
+ + + +
+
+
+ + + +
+ + +
+ +
+ + + +
+ + +
+ + + +
+ + + +
+ + + +
+ + +
+
+
+
+
+ + +
+
+
+ +
+ + + + + +
+ + + + +
+ + + + +
+ + + + + +
+ + + + +
+ + + + +
+ + + + +
+ + + + +
+ + + +
+ + + + +
+ + + + +
+ + + + +
+ + +
+ + +
+ + + +
+ + +
q h hh
+ + +
+ +
+ +
+ + +
+ + +
+ +
+ + +
+ +
+
+ + +
+ +
+ +
+ +
+ +
+ +
+ +
+ + +
+ +
+ +
+ +
+ + +
+ +
+ + +
+ + +
+ + +
+ +
+ + +
+ + +
- 1 z
+ +
+ +
+ +
+ +
+
+
+
+ + +
+ + +
+ + +
+ + +
+
+
+
+
+
+
+
+
+
+
+
vol 3
11. -wun
“-wun
* -wun
' -wulq
-ywun
“-ywun
*-ywun
*-ywulq
12. -en
“-en
' -en
' -elq
-yen
“ -yen
' -yen
■-yelq
-wen
“-wen
* -wen
' -welq
+ + +
+ + +
+ + +
+ + + +
+ + +
+ + +
+ + +
+ + + +
+ + +
+ + + +
+ + + +
+ + +
+ + +
+ +
+ +
+ + +
+ + + +
+ + + +
+ + + +
k kh kk ng
+ +
+ +
+ +
+ +
+ + +
+ +
+ +
+
+ +
+
+ + + + + +
+ + + + + +
+ + + + + +
+ + + + + +
+ +
+ +
+ +
+ +
+ + +
+ + +
+ +
+ + + +
+
+ +
+ +
+ +
+ + + + + +
+ + + 4 - + +
+ + + + + +
+ + + + + +
t th tt n p ph pp m c ch cc s
+ +
+
+ +
+ +
+ +
+
+
+ +
+
+ +
+ +
+ +
+ + +
+ +
+ +
+ +
+ +
+ +
+ +
+ +
ss q h
+
+
+
+
hh
+ + +
+ +
+
+ +
+
+ + +
+ +
+ + +
+ + +
+
+
+
- 1 z
128 PART I
A Reference Grammar of Korean
(vol 3, 12)
/e
kh
kk
ng
t
th tt n
p ph
pp
m
c
ch
cc
s
ss
<1
h
hh
-
1
z
-ywen
+
“f
+
+
+
+
+
+
+
+
+
+
“-ywen
+
+
+
+
+
+
+
+
+
+
+
+
+
+
*-ywen
+
+
+
+
+
+
+
+
+
+
+
+
*-ywelq
+
+
+
+
+
+
+
+
+
+
+
+
+
13. -om
+
+
+
+
"-om
+
* -om
+
+
* -op
+
+
-um
+
+
+
+
+
+
+
"-um
+
+
+
+
+
+
+
' -um
+
+
+
+
+
* -up
+
+
+
+
+
+
+
-i/72
+
+
+ +
+
+
+
+
+
+
"-1m
+
+
+
+
+
+
+
+
* -i/72
+
+
+ +
+
+
+
+
+
+
’ “ip
+
+
+
+
+
+
+
+
+
-a /77
+
+
+
+
+
+ +
+
+
+
+
+
+
+
+
"-am
+
+
+
+
+
+
+
+
+
+
+
+
+
+
* -a/77
+
+
+
+
+
+
+
+
+
+
+
+
* -ap
+
+
+
+
+ +
+
+
+
+
+
+
+
+
vol 4
15. -em
+
+
+
+
"-em
+
+
+
+
+
+
+
' -em
+
+
+
+
+
+
+
•-ep
+
+
+
+
+
+
+
+
-em
+
+
+
+
+ + +
+
+
+
+
+
+
+
+
+
+
+
“-em
+
+
+
V +
+
+
+
+
+
+
+
+
+
+
' -em
+
+
+
+
+
+
+
+
+
+
+
+
+
’ -ep
+
+
+
+
+
+
V +
+
+
+
+
+
+
+
+
+
+
+
16. -woW
+
+
+
+
+ + +
+
+
+
+
+
+
+
+
+
+
+
+
+
“ -wcW
+
+
+
V +
+
+
+
+
+
+
+
+
+
+
+
' -wcW
+
+
+
+
+
V +
+
+
+
+
+
+
+
+
+
+
-ywoW
+
+
+
+
+
+ + +
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
“-ywoW
+
+
+
+
+
V +
+
+
+
+
+
+
+
+
+
+
+
+
+
+
'-ywoW
+
+
+
+
+
+
V +
+
+
+
+
+
+
+
+
+
+
+
+
+
+
17. -wuW
+
+
+
+
+
+ +
+
+
+
+
+
+
+
+
+
+
+
“-wuW
+
+
+
+
+
+
V
+
+
+
+
+
+
+
+
+
+
+
' -wuW
+
+
+
+
+
+
V
+
+
+
+
+
+
+
+
+
+
+
-ywuW
+
+
+
+ + +
+
+
+
+
+
+
+
+
+
+
+
+
+
+
” -ywuW
+
+
+
*- +
+
+
+
+
+
+
+
+
+
+
' -ywuW
+
+
*- +
+
+
+
+
+
+
+
+
+
+
vol 5
18. -o
+
+
+
+
+
**— o
+
+
+
+
■ -o
+
+
+
+
+
-i
+
+
+ + +
+
+
+
+
+
+
+
+
+
+
+
+
+
"-1
+
+
+
f +
+
+
+
+
+
+
+
+
+
+
+
+
* -i
+
+
+
♦- +
+
+
+
+
+
+
+
+
+
+
+
+
-oy
+ +
+
+
+
+
+
+
+
+
+
+
+
+
~-oy
f
+
+
+
+
+
+
+
+
+
* -oy
+
+
»-
+
+
+
+
+
+
+
+
+
+
+
k
kh
kk
ng
t
th tt n
p ph
pp
m
c
ch
cc
s
ss
<1
h
hh
-
1
z
A Reference Grammar of Korean
PART I 129
(vol 5, 18)
-uy
~-uy
■ -uy
19. -woy
"-woy
‘ -woy
20. -ay
“-ay
' -ay
-way
~-way
‘ -way
21. -wuy
"-wuy
' -wuy
-ywuy
"-ywuy
'-ywuy
k kh kk ng t th tt n p ph pp m c ch cc s ss q h hh -
vol 6
2 2. -yey
~-yey
• - yey
-ywey
" -ywey
’-ywey
23. -wo
+ + + + +
2 5. -wu
-ywu
" -ywu
' -ywu
26. -e
k kh kk ng t th tt n p ph pp m c ch cc s ss q h hh - 1 z
130 PART I
A Reference Grammar of Korean
5.0. Forms: nouns.
The noun typically enjoys a certain independence. Unlike verb stems, which require that some
ending be attached, a noun may appear unaccompanied by a particle or other marker. In the broadest
sense, the noun is a kind of default category comprising many subcategories which are defined by
combinatorial restrictions. A pure noun typically can occur as a nominative-marked subject and/or as
an accusative-marked object, while a pure adverb does not attach a case marker. But a word that names
a time or a place functions sometimes as a pure noun and sometimes as an adverb, and there are other
cases where a single word is described as belonging to two or more parts of speech.
5.1. One-shape and two-shape elements.
Certain particles have one shape after a word ending in a consonant and a different shape after a
word ending in a vowel. There are certain inflectional endings, too, which have one shape when
attached to a stem with a final consonant and a different shape when attached to a stem with a final
vowel. (For the purpose of defining vowel-stem and consonant-stem conjugations, the basic -w- of our
analysis, in origin a lenited p, counts as a consonant, and the basic --1-, originally part of the stem, is
treated as an extension of the vowel.) There are other particles and endings, some of which have but
one shape and some of which have more than one but do not select the shape on the basis of the final
phoneme of the element to which they are attached. Particles and endings of the first type can be called
two-shape, and those of the second type can be called one-shape, even though elements of either
type may surface in additional shapes due to other factors, such as the automatic reinforcement of a
voiceless obstruent after a voiceless stop (§2.6). In colloquial Korean the stem of the copula belongs
with the two-shape elements, but our Romanization takes the zero shape as an abbreviation: tangsin ita
‘it’s you’, na ’ta ‘it’s me’. Examples of one-shape particles are ey ‘to, at’, uy ‘of’ (for which the
pronunciation is /ey/, too), kkaci ‘all the way up to, even’, to ‘even, also’, se ‘at, from’, man ‘just,
exactly’, mata ‘each, every’, puthe ‘(starting) from’.Here are examples of two-shape particles,
with the postconsonantal shape (as elsewhere in this book) cited first: i/ka (nominative), ul/lul
(accusative), iya/ya ‘only if it be’, un/nun (subdued focus), kwa/wa ‘with, and’, iyo/yo (polite
style). The particle ulo/lo (manner, direction, state or change of state, means, reason, ... ) is peculiar
in that the postvocalic form lo is used also after the consonant •••!, as in yenphil lo sse ‘writes it in
pencil’; contrast yenphil ul sse ‘uses a pencil’. Examples of the one-shape and two-shape endings will
be found in §9.
5.2. Nouns.
A noun, in the broad sense, occurs in at least one of four environments:
(1) before a particle: achim i wass.ta ‘morning has come’;
(2) before the copula i- as a complement: achim ita ‘it’s morning’;
(3) before a noun or noun phrase which it modifies: achim hayq-pich ‘morning sunlight’;
(4) in absolute constructions, which may be interpreted in any appropriate role, including
adverbial: achim wass.ta ‘morning has come’ (= achim i wass.ta) or ‘arrived in the morning’ (=
achim ey wass.ta).
In normal speech, nouns are never followed by pause in environments 1 and 2; in environments 3
and 4, pause is more frequent, especially in 4. The English obligatory categories of singular/plural,
definite / indefinite, and general / specific are essentially absent from Korean nouns. Without special
marking, as by a numeral or by an element such as - tul ‘(as) a group’, we are not told whether chayk
means ‘(a) book’, ‘the book’, ‘(some) books’, or ‘the books’. Specific words are intrinsically singular
or plural, notably such pronouns as na ‘I/me’ and wuli ‘we/us’, but the intrinsic meaning may be
overridden by semantic extensions, just as the English royal or editorial “we” is often used as a
singular. According to Seok-Choong Song (Song Sekcwung) “plural marking in Korean individuates,
whereas the unmarked category categorizes its referent”; he notes that plural marking is obligatory for
nouns that have “specific reference”, and that is borne out by pronouns, proper names, and the like.
A Reference Grammar of Korean
PART I 131
5.2.1. Quasi-free nouns.
A quasi-free noun has great freedom of combination but it is always preceded by an element such
as i ‘this’, ku ‘that’, or ce ‘that (yonder)’, or by an adnominalized phrase ‘which (is/does) ... ’ or a
modifying noun phrase. (In contrast, the quasi-adnouns, §5.3.1, are always followed by a noun or a
noun phrase.)
The following list of quasi-free nouns includes some that are often called “imperfect nouns” (pul-
wancen myengsa) or “dependent nouns”. I refer to some of these as postnouns and postmodifiers.
What a given quasi-noun may be preceded by is an individual property of the word. Some of the words
are occasionally free under highly restricted circumstances: thek need not be modified when it is
followed by eps.ta or eps.i. But idiomatic expressions of that sort deserve separate entries in the
lexicon.
LIST OF QUASI-FREE NOUNS
- ca ‘person’
••• ccak (= ccok) ‘direction’;
= kkol ‘appearance’
- ccok ‘direction’
••• cek ‘time’
- chi ‘stuff, thing; guy, one’
••• chuk ‘side’
••• chwuk ‘group’
••• els ‘act, motion’
••• cuum (cium) ‘approximate time’
••• hay ‘possessed thing, one’s’
••• i ‘person’
- ilsswu ‘constant (bad) habit’
? ••• keli ‘material, ... ’
••• kes ‘thing, one, fact, ... ’
••• key ‘one’s place, home’
- kkan (ey) ‘by one’s own account’
••• kos ‘place’
••• mulyep ‘time’
••• nolus ‘job, role’
- ppal ‘manner’ (rare)
••• pun ‘esteemed person’
••• tey ‘place, ... ’
••• thas ‘fault’ (also verbal noun
transitive ‘blame’)
- thek ‘reason, limitation; resources; ... ’
••• ttolay ‘of (that) age or size’
••• ttan (ey) ‘(by) one’s own kind
judgment’
? ••• ttawi ‘of the sort, and the like’
? - tungci ‘vicinity’
- tungtung ‘et cetera’
Restrictions on the occurrence of the quasi-free nouns vary. Some are severely limited, as shown
by the individual entries in Part II. Certain postmodifiers and postnouns that can perhaps be regarded
as quasi-free have not been included here for various reasons. A few words that others have included
in this category are omitted because I have found them in sentence-initial position and decided to treat
them as free nouns: ttaymun ‘reason; the sake of ’ is omitted only because a sentence can begin
Ttaymun ey ••• ‘Therefore ’ and we might well consider that usage an abbreviation of Kuleh.ki
ttaymun ey ••• ‘Because of its being that way ••• ’. See also the entries chwuk, nom, nyen, and nyesek
in Part II. And notice the uncommon use of palo (3 in the entry of Part II).
In contrast with the quasi-free nouns, BOUND NOUNS (§5.2.8) occur only in very limited types of
compound (like the “cran-” of English “cranberry”). Most are here treated as bound adnoun =
prenoun (or prefix), bound postnoun (or suffix), and bound preverb (bound adverb or prefix). There
appear to be several bound preparticles, as listed in §5.2.9.
5.2.2. Free nouns.
A noun that is further unspecified is simply a free noun. At present I do not break the category
down into as many subclasses as might be desirable for various purposes. I fail to distinguish COUNT
nouns, MASS nouns, ABSTRACT nouns, etc., though the distinction can surely be drawn on the basis of
the selection of counters; nor do I here distinguish between animate and inanimate, but see §10.8.8
for a useful correlation of distribution with the corresponding verb classes, in that there are verbs that
have only animate subjects or only animate objects (or both). A more refined classification will emerge
132 PART I
A Reference Grammar of Korean
from farther syntactic analysis. For purely practical convenience, I set up the subclass of PROPER
NOUNS and that of deictics, which includes pronouns and indeterminates (interrogative-indefinites).
There are also deictic verbs and adjectives (kule- ‘do that way’, kuleh- ‘be that way’). Verbs of motion
can also be described as deictic, since the choice of o- ‘come [to where I am]’ and ka- ‘go’ depends on
the position (location or psychological involvement) of the speaker and the hearer: o- “move toward
me/here/now” is the semantically marked form, and ka- is the default. On the subtleties of choosing
one or the other of this pair of verbs, see 'Yi Cenglo 1985. (Standard Japanese uses the corresponding
pair of verbs in a similar fashion. There may be dialects in both Japan and Korea that differ from the
standard usage.) *Yi Kitong 1988 describes the semantic difference between o- and ka- as moving
toward or away FROM the “deictic center”, which includes not only “me/us, here, now” but also
“the normal / desirable state, proper shape/conditions”.
5.2.3. Proper nouns: names and titles.
Personal names are often of two syllables, and one of the syllables is sometimes used as a
generation reference, so that brothers may be named Sengen and Cwuen, or Cengkil and Cengmin.
The Chinese character for the common syllable is called the tollimq ca or hang.lyelq ca. Female
names often take as a second syllable the suffix -huy (a Chinese loan meaning ‘princess’), and in rapid
pronunciation -huy sounds like -i, the common hypercoristic suffix that is added only to names ending
in a consonant and is not to be confused with the nominative marker i/ka: Chwunhyang-i (ka/lul)
‘Little Chwunhyang [as subject/object]’, Chwunhyang (i/lul) ‘Chwunhyang [as subject/object]’,
Chwunhuy (ka/lul) ‘Chwunhuy [as subject/object]’. A girl who is called /swuni/ may bear the
straightforward name Swunhuy, incorporating the suffix -huy, but she may write her name with a
spelling variant Swuni; or, she may write it Swun-i, especially when it is a short form for names like
Pokswun(-i) or Cengswun(-i). Similar short forms are heard for other names, such as Swuk-i for
Yengswuk-i, Tong-i for Poktong-i, ... . Either the family name or the given name may be of one or
two syllables - rarely, even three - so that a full name may have as few as two or as many as four
syllables: He Wung, Payk 'Nakcwun, Sen.wu Wung, Ulqci Muntek, He Nanselhen. The three-
syllable type is the most common in Korea, as it is in China. Given names of three syllables are
unusual and typically of non-Chinese origin, though the syllables may sound like Chinese elements,
and characters are sometimes assigned to them on a phonetic basis. Korean names are usually said with
the family name first, as in Chinese, Japanese, and Vietnamese. When the personal name has two
syllables, some people like to Romanize it as two words, with or without a hyphen. A man named Kim
Cen.il, for example, may want to write his name as Kim Cen II or Kim Cen-Il and in English call
himself C.I. Kim. Some Koreans have “foreign” names, either in addition to Korean names, or in
place of them; the name is often Biblical. When the family name is Korean it is probably better to write
that first: Kim Phollin (= Pauline), Cen Tawis (= David). If the name is that of a foreigner (other
than Chinese or Japanese), it is best to leave the name in the foreign order even in Hankul, and to let it
revert to the foreign spelling when Romanizing: “Samuel E. Martin” not “Martin Samuel-E.”. In
English texts Koreans usually try to follow the foreign order, so that the late 'Yi Sungman is referred
to by the English spelling he preferred “Syngman Rhee” and the educator Payk 'Nakcwun writes in
English under the name “George L. Paik”.
A name, either a surname or a full name, is often followed by a TITLE: sensayng (or with varying
connotations ssi, kwun, sepang, ...) ‘Mr’ (ssi can sometimes mean ‘Mrs’ according to Roth 281), puin
or stylishly samo (nim) or sensayng (nim) samo or sensayng samo nim ‘Mrs’, yang ‘Miss’,
sayngwen ‘Mr’ or ‘ ••• Esquire’, cwusa ‘director; petty officer’, paksa ‘Dr’ (Ph.D.), uysa ‘Dr’
(medical), moksa ‘Reverend’, sinpu ‘Father’, kyoswu ‘Professor’, kak.ha ‘His/Your Excellency’,
cenha ‘His/Your Royal Highness (Prince - )’, sengha ‘His/Your Holiness (the Pope)’, sengsang
‘His/Your Majesty (the King)’, phyeyha ‘His/Your Imperial Majesty’, etc. Most of the titles can be
followed by the postnoun nim ‘esteemed’, and nim sometimes follows a name directly: Yeyswu nim
‘Jesus’, Sek.ka-yelay nim ‘Buddha’, Kongca nim ‘Confucius’, Mayngca nim ‘Mencius’. In bookish
style contemporary names occur this way, too: Kim nim ‘(Mr) Kim’, Sen.wu nim ‘(Mr) Sen.wu’.
A Reference Grammar of Korean
PART I 133
There is an adnoun title Seng ••• ‘Saint ••• In addition, two adnouns are in current use as titles in
South Korea: Misu ‘Miss’ (of a young unmarried professional woman or office colleague), Misuthe
‘Mr’ (of a male colleague): wuli cicem ey se il hanun Misu Kim hakwu Misuthe Pak un ‘Miss Kim
and Mr Pak who work in our branch office’. In North Korea, the postnoun title tongci ‘comrade’ is in
vogue: Kim tongci ‘Comrade Kim’. Placenames have a basic form (like Phyengyang, Payktwu,
^aktong, Ceycwu, etc.) which can be considered a free noun, even though frequently the more
common version includes a following category designator to specify the kind of place: Phyengyang si
(‘city’), Payktwu san (‘mountain’), 'Naktong kang (‘river’), Ceycwu to (‘island’). Other common
designators are to ‘province’, kwun ‘county (prefecture)’, myen ‘township’, and the somewhat less
productive Velqto ‘archipelago’, (-)man ‘bay’, (-)hay ‘sea’, (-)yang ‘ocean’, (-)hang ‘port’, (-)ll
‘village’, (-)kwu ‘ward’, (-)kwung ‘palace’, (-)sa ‘temple’, (-)yek ‘station’, (-)sen ‘line’, (-)to ‘ferry’.
And to the list we can add (-)mun ‘gate’ as in Kwanghwa mun (the main gate of Kyengpok Palace),
(-)cwa ‘constellation’ as in Kolay cwa ‘the constellation Cetus’. As remarked earlier, Koreans show a
marked tendency to take any monosyllabic word and tack it on to an adjacent word, so that a name like
Nam san ‘South Mountain’ or Han kang ‘Han River’ is often taken to be a single word. For
consistency, it seems better to Romanize the more productive category designators as separate words,
even though they are monosyllables and even when they are attached to other monosyllables. Notice
that to ‘island’, to ‘ferry (point)’, and often (§2.7.2) to ‘province’ are homonyms. Some of the
provinces are divided into North and South (like the Dakotas): Phyengan puk to ‘North Phyengan
(province)’, Phyengan nam to ‘South Phyengan (province)’. When writing, Koreans often treat puk-to
and nam-to as units. A list of the Korean provinces will be found in Appendix 3. In Appendix 4 you
will find lists of Japanese placenames with their Korean readings. It must be kept in mind, however,
that Koreans often use phonetic approximations to the Japanese pronunciation for many Japanese
names, and especially for those which are not of Chinese origin, such as Nagasaki (Nakasakhi). For
other foreign names there are sometimes two forms: Mikwuk or Ameylikha ‘America’. In general the
“foreign” forms are more modern and sophisticated, but those based on Chinese characters are more
succinct and often better known. (They are also easy to abbreviate: Mi-Han - ‘American-Korean ’.)
Other proper names are book titles, corporation names, and the like. These are often characterized
by abbreviation and ambiguity, sometimes intentional, so that it is not always easy to figure out the
appropriate word division for the Romanized form. Such a proper noun will frequently have a final
category designator that functions like those for places mentioned above: (-)sa ‘company’, (-)sa
‘history’, (-)cen ‘tale’ or ‘biography’, etc.
5.2.4. Deictics.
Deictics are those elements which alternate in reference depending on who is speaking. To be
consistent we would have to include as deictics the honorific marker -(u)si-, which marks the subject of
a verb as someone other than the speaker because it is someone toward whom the speaker is showing
special esteem; the honorific particle kkey ‘to someone esteemed’; personal names; and perhaps a few
other things that, for various reasons, we will treat separately. Notice that names and titles are very
often used as pronominal substitutes. Perhaps the most common polite way to say ‘you’ is sensayng
nim or sensayng. Without going into all the details of usage, we can assemble the following lists.
PRONOUNS
I/me
na (/nay)
we/us
wuli
I/me [formal]
ce (/cey)
we / us [formal]
ce-huy
you
ne (/ney)
you all
ne-huy
you [familiar]
kutay
you all [familiar]
ku (ney) tul
caney
caney tul / kkili
i salam 1
i(-i) tul/kkili
you [impersonal]
tangsin
you all [impersonal]
tangsin tul / kkili
134 PART 1
A Reference Grammar of Korean
you [to inferior]
ce (/cey)
you all [to inferiors]
ce tul/kkili
you all [disrespectful]
keyney 2
you [formal]
sensayng (nim)
you all [formal]
yele pun
‘Sir’
sensayng (nim)
‘Madam, Ma’am’
puin / samo (nim)
oneself
caki, ce,
casin, cachey
themselves
caki tul, cehuy
itself
cachey
he/him, she/her
ku - 0, salam,
they/them
ku - tul,
pun, nom, ca, ...)
ku - ney (tul)
ce - (i, salam,
ce - tul,
pun, nom, ca, ...)
ce - ney (tul)
it (‘this’)
i (kes)
they/them
i (kes) tul
(‘that’)
ku (kes)
ku (kes) tul
(‘that’)
ce (kes)
ce (kes) tul
1 Also: ‘my spouse (he/she)’. 2 ?< keki ney; ?< ku ai ney (Cf kaytul < ku ay tul)
CORRELATIVES
Indeterminate
Generic Proximal
Mesial
Distal
enu
amu
i/yo
ku/ko
ce/co
‘which, some’
‘any(one)’ ‘this (one)’
‘that (one)’
‘that (one) yonder’
eti
(amu tey) yeki / yoki
keki / koki
ceki / coki
‘what / some place’
‘any place’ ‘this place’
‘that place’
‘that place yonder’
ecci
amuli
ili / yoli
kuli / koli
celi / coli
‘what/some way’
‘any way’ ‘this way’
‘that way’
‘that way yonder’
ette (ha-)
amule
ile/yole
kule / kole
cele/cole
‘how, somehow’
‘anyhow’ ‘thus’
‘so’
‘so, yea’
Notice that eccay = ecc’ ’ay is contracted from ecci hay ‘is/does what way’, and etteh-,
amuleh-, ileh-, kuleh-, etc., are from ette ha-, amule ha-, ile ha-, kule ha-, etc.
The word eti can be regarded as from etey, a dialect variant of enu tey ‘what place’ (Cf p. 25).
And encey ‘what/some time, when’ is contracted from enu cek (ey) ‘(at) what time’; it is unclear just
how the related words i(n)cey ‘now’, ecey ‘yesterday’, and kucey /kucekkey ‘day before yesterday’
were derived. Other indeterminates are nwukwu (but nwu before the nominative marker ka)
‘who/someone’, mue(s) ‘what/something’ (obsolete musum/musam), musun ‘which/some - ’,
meych < myech ‘how much/many, some amount/number’, elma ‘what/some quantity’ (< "erima
< *e'nu 'ma), and way ‘for what/some reason, why’, which has the shape weyn when adnominal, as
in weynq II in ya ‘what’s the matter?’ (with falling intonation) or ‘is something the matter?’ (with
rising intonation). The word etten ••• ‘what sort of ••• ’ is the modifier form of etteh-, thus ultimately an
abbreviation of ette han ••• ‘ ••• that is what way’. Although I am unable to offer examples of early
sentences that use the indeterminates in a non-interrogative way, I presume that the language of the
15th century did not differ from the modern language in that respect.
There are a few paradigmatic gaps in the use of the deictics. As Cang Sekcin observed, you can
say i ttay ‘this time’ and ku ttay ‘that time’ but not *ce ttay ‘that [distant] time’. The derived forms
ipttay ‘up to now’ (= yethay) and cepttay ‘not long ago’ are not paralleled by *kupttay ‘up till then’.
The proximal and mesial deictics can also be used anaphorically, but not the distal. In that respect,
Korean differs from Japanese, where the corresponding (k]a- words can be used to mean ‘that obvious
A Reference Grammar of Korean
PART I 135
■■■ [known to both you and me or to all]’. For Korean anaphora there is only a two-way distinction of i
and ku (Pak Hwaca 1982). In i kes ce kes ‘this and that; something or other’ the distal deictic is not
anaphoric though it is indeed metaphoric, for the ‘that’ is not visible.
The connotations of personal pronouns are apt to change through time. The anaphoric designator
ku ‘that one’ is used as a third-person pronoun only in rather formal writing, for it is impersonal as
compared with ku salam/i/nom/... . When used, it has a masculine orientation, but it can also refer
to females. A fairly new (post-1945?) pronoun ku-nye (perhaps modeled on Japanese ka n no-zyo) is
used consistently by some authors for references to ‘she/her’ while others refer to females by using ku
and ku-nye interchangeably. 1880 Underwood says i, ku, and ce are “disrespectful when referring to
people”. These days it is quite popular to use terms with the honorific - nim for pronominal reference.
Intimates sometimes use caki ‘oneself to refer to either the first or the second person. For the second
person, polite usage calls for a title or name + title: sensayng nim, Kim sensayng ‘you(, Mr Kim)’.
Informal words for ‘you’ include i salam, which can also be used to mean ‘this person; he/him, she/
her’. And i phyen can refer either to ‘you (all)’ or (= i ccok) to ‘I/me; we/us’, in addition to the
basic meaning of ‘this side’.
5.2.5. Adverbs.
An adverb is a noun that occurs typically (and a few of them perhaps exclusively) in absolute
position, i.e. as an adverbial phrase. There are also unusual cases where the adverbial phrase
modifies an entire copula sentence: Pelsse chwulkun ip.nikka ‘Are you leaving for work already?’;
Pelsse Taycen ita ‘It’s already Taycen’ (on a train trip); Tayk i palo Kim Pok.il ssi ’sey yo? ‘Are
YOU, then, Kim Pok.il?’; Enu-tes kaul iess.ta ‘It was autumn before we knew it’; Acwu yatan tul ita
‘What a fuss!’ (CM 2:52); Kkok machan-kaci ’ta ‘They are exactly the same’ (CM 2:52); I ttang un
on-thong tol-path iess.ta ‘This land was all a field of stones’ (CM 2:52).
One step removed is the still more unusual case where the adverbial phrase is adnominalized
(§11.9) by position only: kas sumul (ita) ‘(he is) just 20’, palo ku chayk ‘that very book’; tan hana
(lul) ‘just one’, kyewu twul (in ya) ‘only two?’ This is especially common with time and place nouns:
te alay ‘farther down’, cokum aph ‘a little ahead’, acwu choykun (ey) ‘quite recently’. One case is
especially interesting: Kkway yele chayk tul i iss.ta ‘There are quite a lot of books’, in which the
adverb kkway apparently modifies the adnoun yele despite Kkway chayk i iss.ta ‘(he) has quite a lot
of books’, in which kkway modifies the sentence chayk i iss.ta ‘books exist’, because we can say
Kkway yeles i iss.ta ‘There are quite a lot of them’. Cf (CM 2:56) kkway say kenmul ‘quite a new
building’, acwu yele saqken tul ‘very many incidents’. (For a somewhat different interpretation of
these structures, see CM 1:453-4.) Adverbs, especially those of degree, can modify other adverbs:
acwu ppalli ‘very quickly’.
Most adverbs can be followed by either the particle un / nun or the particle to, and the ubiquitous
particle tul sometimes attaches to an adverb: Phyen hi tul hasey yo ‘Take it easy (you people)!’ The
only clear exceptions seem to be mos ‘definitely not, cannot’ and an(i) ‘not’, and certain conjunctional
adverbs (mich, cuk, ko lo, ... ); KEd carries one example of an tul V: An tul mek.nun ya? ‘Aren’t
you folks eating?’ It has even been suggested that, after all, to may just be possible with both an(i) and
mos as these examples indicate: Ku nun kongpu lul ani to halye V i wa meli to napputa ‘On top of
not studying, he has a poor head, too’; Ku nun kongpu lul mos to halye V i wa nung.lyek to eps.ta
‘In addition to not studying, he is lacking in ability, too’; Cham mos to sayngkyess.ta ‘How ugly he
is!’. In Middle Korean a'ni could be followed by 'two and non and sometimes an adverb or adverbial
phrase intervened before the verb; see Part II. What is more, a'ni can appear directly before the
accusative, locative, and comitative markers as if it were a noun. Such structures are the result of a
direct nominalization of a noun predication with the copula form (we expect i'lwom or ’i'ywom)
ellipted and the particles attached to a 'ni itself:
"twulfhj a'ni lol cwo'cha 'ssywun 'hosya'm i'la (1465 Wen 1:1:2:57a) ‘it is that he follows
pursuing what is not two [but one]’.
136 PART I
A Reference Grammar of Korean
"twul'h i "twulfti] a'ni lol U'hwu'm i khwong-syang i'la 'ho'si.n i ’’la (1464 Simkyeng 38a)
‘said the name for two not being two is “unreality”’
swon s kalak 'kwa swon s kalak a'ni yey "na'mo'n i 'Gwa a'ni 'Gway "twul'hi "ep.su'mol
nilo'si.n i ’ la (1462 'Nung 2:61b) ‘said that to the finger and the non-finger, the remaining and the
non-remaining there are no two of them’.
Ani is used alone before the copula (Ani ’ta ‘It is not’) and as an interjection meaning ‘No!’; mos
is sometimes followed by the versatile particle tul. The only other occurrence of the morpheme mos
not directly followed by a verb seems to be in the word calmos ‘mistake’, derived from cal mos -
‘can’t ~ very well’; calmos is also used in absolute position, as an adverb, so that calmos hayss.ta can
mean either ‘made a mistake’ (= calmos ul hayss.ta) or ‘did it wrong’ (= - calmos hayss.ta). Cf cal
(I) mos hayss.ta ‘did (or could) not do it well’
Among the other adverbs those which seldom, if ever, occur with focus particles such as un / nun
or to, can usually be followed by the plural marker tul, which is the ultimate test for separability.
Examples of adverbs marked by un/nun, to, man, etc., will be found under the entries for individual
particles in Part II. Most adverbs of time can also take the ablative puthe ‘(starting) from’ and/or the
allative kkaci ‘(continuing) all the time till’: Ilccik puthe al.ess.ta ‘I knew it from early on’; Akka
puthe kitalyess.ta ‘I’ve been waiting for some time; I started waiting a while ago’. Cf pelsse puthe
‘for some time now’.
If we were to regard the adverb as a noun that has dropped its marker (a handy but inaccurate
concept), the appropriate particle would be ulo/lo or, especially with time words, ey. Sometimes we
find parallel or competing expressions, with and without the particle: pothong (ulo) ‘usually’, onul
(ey) ‘today’, ili (lo) ‘this way’.
The following lists are not exhaustive, but ample. To make the lists more useful, I have divided
the adverbs into rough semantic categories, in lieu of the more rigorous groups that will have to await
further study: (1) adverbs of time, (2) adverbs of degree, (3) adverbs of contingency, (4) adverbs of
assertion, (5) conjunctional (connective) adverbs, (6) adverbs of manner. There is overlap among the
lists and with other lists; way, for example, is also listed as a deictic. Usually the English translations
are enough to indicate which items are used frequently in other than absolute position (e.g. onul
‘today’). With a few exceptions, I have included neither phrases nor the large number of derived
adverbs such as ppalli ‘fast’ and the phrases with - hi (< ha-). Notice also the regular inflectional
category -key called adverbative (§9).
(1) ADVERBS OF TIME
Many of these words are pure nouns that are directly adverbialized. We know they are pure nouns
because they can be used as subjects and objects: Cikum i palo nala lul wi hay se II ul hay ya halq
ttay ’ta ‘Now is the time we should work for the nation’. But some of the words in the list (kot
‘immediately’, pelsse ‘already’, ...) are not pure nouns, for they cannot be so used.
encey ‘when; sometime’
(t)tayttum ‘at once’
mak ‘just (at the moment); just now’
pelsse ‘already’
pelsse puthe ‘for some time now’
imi ‘already’
icey, incey ‘now; from now on’
cikum ‘now’
sipang [? lit] ‘now’
pangkum ‘just now (= a bit ago)’
kumpang ‘just now (= shortly)’
tangkum ‘at present’
tangpun-kan ‘for the time being’
say lo ‘newly’
kas ‘just ( + ages by tens), barely, freshly,
newly (born)’
onul (nal) ‘today’
1 nayil ‘tomorrow’
ec(ekk)ey ‘yesterday’
kuc(ekk)ey ‘day before yesterday’ (Kyengsang
dialect alay)
ku-kkuc(ekk)ey, samcak-il ‘three days ago’
A Reference Grammar of Korean
PART I 137
moley ‘day after tomorrow’
kulphi ‘three days from now, the day after the
day after tomorrow’
ku-kulphi ‘four days from now’
kumnyen ‘this year’
cak.nyen, clnan hay ‘last year’
kulekkey, caycak.nyen, ci-cinan hay ‘year
before last’
ku-kkulekkey, samcak-nyen ‘three years ago,
the year before the year before last’
'naynyen, myengnyen ‘next year’
‘nay^ay-nyen, hwunyen, caymyeng-nyen
‘year after next’
'nay-hwunyen, hwu-hwunyen ‘three years
from now’
hwuq-nal ‘someday (in the future)’
cangcha ‘in the future’
aph ulo ‘in the future’
ilkan ‘in a few days’
taum ‘next’
chacha, chachum ‘gradually’
cemcha lo ‘gradually’
cemcem ‘more and more, gradually’
tangcho’y = tangcho (ey) ‘at first,
originally’
ponti, wen.lay, wenak, wenchey, a.ye(y)
‘from the beginning’
ponsi ‘originally, formerly’
nul, hangsang, hangyong ‘always’
nosang ‘constantly’
‘yensok ‘continually’
cwul-kot ‘continually’
yeng(yeng) ‘forever’
cina-sayna ‘night and day’ (< clna sayna)
cacwu ‘often’ (< cac-wu, derived adverb)
mayil ‘every day’
maypen, mayyang ‘every time’
maywel ‘every month’ (= tal mata)
maynyen ‘every year’ (= hay mata)
camsi, camqkan ‘for a little while’
olay ‘(for) a long time’ (< adjective infinitive)
twuko-twuko ‘for a long time’ (< vt gerund)
tangcang ‘then and there, on the spot’
kot ‘immediately’; ‘id est’ -» (5)
kot-cang ‘right away’
samus ‘right away’; ‘quite (different)’ -*■ (2)
inay ‘immediately (after)’
elph u /js ‘at once’
nallay, nayngkhum / nuyngkhum ‘promptly,
lickety-split’
phettuk ‘in a flash’
enu-tes ‘before one realizes it, in no time’
enttus ‘suddenly (seeing)’ (? < enu-tes)
chenchen hi ‘slowly’
ppalli ‘quickly’ (derived adverb < ppalu-)
ellun ‘at once’
ese ‘right away’; ‘please’ -* (3)
akka ‘a (little) while ago’
coman-kan ‘sooner or later’
iss.ta (ka) ‘after a while, presently, shortly’
elma an ka (se) ‘soon, before long’
me(l)ci anh.e ‘soon, before long’
kumsay = kumsey (< kumsi ey) ‘any minute
(now)’
mikwu ey ‘shortly, soon’
mence ‘first of all’
mili ‘in advance’
ciley ‘in advance, beforehand’
piloso ‘initially’
cheum (ulo) ‘for the first time’
ilccik ‘early’
ilqtan ‘(when) once; for the moment’
*imsi (lo) ‘temporarily’
(kkuth-)kkuth-nay ‘to the last, to the end’
olay kan man ey ‘at long last’
mo chelem ‘at long last; with great effort’
tutie, machim-nay ‘at last’
nacwung (ey) ‘finally’
ttay-ttay lo ‘occasionally; now and then’
kakkum ‘occasionally’
ittakum ‘occasionally, now and then’ (<
iss.ta + -kum)
tele ‘occasionally’; ‘somewhat’ -*• (2)
com chelem + negative ‘seldom’
(ttay) machim ‘just in time’
twl-miche ‘soon after’
miche ‘(not) yet’ (< michye, vi infinitive)
acik, acik to ‘still, yet’
han-kkep(en) ey, tan-swum ey, tan-khal ey
‘at one time, at a stretch’
like ey ‘at one stroke, at the same time’
hamkkey ‘together’
kath.i ‘together’; ‘like’ -* (6), derived
adverb < adjective kath-
iuk.ko ‘in a short while’, abbr < iuk
(= isuk) hako
138 PART I
A Reference Grammar of Korean
(2) ADVERBS OF DEGREE (and QUANTITY)
A few of these words (ta, motwu, ... ) also function as pure nouns and can be marked as subject
(••• i/ka) or object (••• ul/lul).
phek, phek una ‘very’
kkway ‘very; relatively; fairly well’
ssek ‘very’
ocwuk ‘very; indeed’
maywu ‘very’ (derived adv < adj mayw-)
taytan hi ‘very’ (derived adverb)
slm hi ‘(does) very much’ (derived adverb)
(ci)kuk hi ‘very’ (derived adverb)
nung hi ( - halq swu iss.ta) ‘nicely, easily
(able to do)’ (derived adverb)
kkumccik ’i ‘exceedingly’ (derived adverb)
hako ‘muchly’ (gerund of obsolescent adj ha-
‘great; much, many’)
ha to ‘indeed’ (infinitive + particle)
sangtang hi ‘relatively, comparatively,
rather’ (derived adverb)
Cham ‘real, very’; ‘truly’ -* (4)
yekan ‘to (no) usual degree’
cak-cak ‘moderately, not too much’
ta ‘all’
coy(-ta) ‘all’
motwu(-ta) ‘all’ < mot-wu, derived adverb
< vt mot- (+ adv ta)
mot’-ta ‘all’ (abbreviation of motwu-ta)
kac.chwu(-kac.chwu) ‘all’ (derived adverb
< vt kac.chwu-)
(on-)thong ‘totally, completely, all’
mocoli ‘entirely’
cen-hye ‘totally’ (< cen ha.ye)
yeng ‘totally, completely, quite’ (NKd
4628b); (= yengyeng) ‘forever’
kkang-kuli ‘wholly’ (derived adverb < vt
kkang kuli- ‘finish’)
nemu ‘too, overly, to excess’ [+ adj or verb]
(< nem-wu, derived adverb < vi nem-)
humssin, humppek ‘thoroughly, to the fullest
(measure)’
cinthang ‘to one’s fill’
tampp a / u k, tempp e / u k ‘overflowing’
hansa kho ‘to the bitter end’
oloci ‘mainly’
haphil ‘of all things (persons, places)’
kocak ‘at most’ (= manh.e to)
ceng ‘very’; ‘truly’ -*■ (4)
mopsi ‘terribly’
acwu ‘extremely; quite’
te ‘more’
tewuk, tewuk(.)i, tewuk te ‘still more’
com te ‘a little more’
han-kyel ‘much more; especially’
han-chung ‘all the more’
tel ‘less’
tele ‘somewhat’; ‘occasionally’ -* (1)
ke(ci-)pan ‘over half, nearly (all)’
yepuk ‘how very much’
hwelssin by far, overwhelmingly’
kacang ‘most; very’
camos ‘highly, exceedingly’
samus ‘completely (different); ? very hard’;
‘right away’ -*■ (1)
muchek ‘exceedingly’
mulye (+ NUMBER) ‘no more than ’; Cf
pulkwa (adnoun, p. 148)
mulus (+ NUMBER) ‘ - or so’
taman ‘just’
ocik ‘only’
tanci ‘merely’
cokum, com ‘a little’; ‘please’ -» (4)
ceypep ‘fairly, tolerably, passably’
yak.kan ‘some; somewhat’
keuy, kecin ‘almost’
kyewu ‘hardly’ (derived adverb < - kyew-)
kansin hi ‘barely’
pyel lo ‘(not) particularly’
(3) ADVERBS OF CONTINGENCY
man.il < "man'il < men- qilq ‘if’ (— ...
-umyen,-ess.tula’myen; ~ -kyengwu eynun)
man .yak < men-ZYAK ‘if (perchance)’
sellyeng, selhok, selqsa ‘if (mayhap)’
kalyeng, kasa ‘if (say)’
pilok < pi'lwok ‘even if’ (~ —ci man, -e to,
-ulq ci ’la to, -una, -kena, -ko to; — - ila to)
am’ man ‘however (it may be)’
amuli ‘however (much)’
machi ‘like, as, if’
(4) ADVERBS OF ASSERTION
ama (to) ‘perhaps; likely, probably’
kulssey ‘maybe; well ... ’
ha.ye-kan ‘anyway, at any rate’
hamulmye ‘all the more/less (so)’
molumciki ‘by all means, necessarily’ (<
molum cik hi, derived adverb)
A Reference Grammar of Korean
PART I 139
ha.ye-thun ‘anyway, at any rate’ (< ha.ye
hatun)
kwa.yen, kwasi ‘sure enough’
ttan un ‘now that I come to think of it, as for
that, to be sure’
cengnyeng ‘(for) sure, certain(ly)’
(cin)sil lo, sil un ‘truly’
ceng ‘truly’; ‘indeed, very’ -* (2)
cham (ulo) ‘truly’; ‘very’ -* (2)
ceng-mal ‘truly’
cinceng (ulo) ‘really’
cham-mal ‘truly’
sasil (lo/un) ‘in truth’
ccacang ‘for a fact, truly’
mlsangpul ‘indeed’
ttok ‘exactly’
palo ‘right, precisely’
cuk ‘precisely’; ‘id est’ -* (5)
an(i) ‘not’
mos ‘definitely not; cannot’
toce-hi ‘cannot possibly’
selma (< hyelma < hyen-ma, Cf elma)
‘by no means, surely (not)’
ilwu ‘(cannot) possibly’ (derived adverb <
ilu- ‘reach’)
kyel kho ‘absolutely (not)’ (< ••• hako)
cokum to ‘(not) even a little’
(5) CONJUNCTIONAL (CONNECTIVE) ADVERBS
mich [literary] ‘and’ (also particle)
nayci (nun) [literary] ‘and’
tto-nun ‘and/or’
tto ‘and, moreover, further’
tto-han ‘too, also, as well; either’
yeksi ‘too, also, as well’
hok(-un) ‘or else, or again’
thel-kkuth mankhum to ‘(not) even a bit (a
“hair-end”)’
tom(uci) + negative ‘not at all’
hwaksil hi ‘certainly’ (derived adverb)
mattang hi ‘by rights’ (derived adverb)
ttopaki ‘without fail; completely’
mullon, mulon ‘of course’
pitan + NEGATIVE ‘not only’
kwuthay(e) ‘[not] making a special effort
(troubling oneself to do so)’ [usually spelled
kwuthayye]
ceypal ‘please, hopefully, I hope’
puti ‘please’
ese ‘please’; ‘right away’ -» (1)
amu-ccolok ‘by all means; please’ (? < amu
ecci ha-tolok)
u y lyey (hi / lo) ‘ as usual; as a matter of
course; without fail; for sure’
ungtang ‘for sure’
kkok ‘for sure’
pantusi ‘certainly’ (? < pantus ’i ‘straight’)
kiphil kho ‘by any means, at all costs’
kie-(h)i ‘by all means’
tay-kwancel, to-taychey ‘(wh-) on earth?!’
taypem ‘on the whole, in general’
taykay ‘on the whole, in general’
mulus ‘on the whole, in general’
tasi ‘or, again, further’
kot ‘id est’; ‘immediately’ -* (1)
cuk ‘id est, namely, to wit’; ‘precisely’ -* (4)
tekwu(nta)na ‘moreover, in addition’
tolie ‘instead, rather, on the contrary’
ohilye ‘rather, preferably, more likely’
ko lo ‘therefore, so’
+ various forms of kule (kuli) ha-, etc., such as kulemyen ‘then’, kuliko ‘and’, etc.
+ “sentence connectors” (§13)
Note: kyem ‘and, as well as’ is treated as a postnoun.
(6) ADVERBS OF MANNER
ecci ‘how; why’
way ‘why’
cal ‘well, nicely’
tul ‘plurally, as a group’ (also other parts of
speech; see Part II)
kak-kak ‘each, every; respectively’;
— ulo ‘from moment to moment’
cey-kak.ki, cey-kakkum ‘each one,
severally’
selo ‘mutually; together’
ta-cca ko-cca lo ‘without the least warning,
unexpectedly, suddenly, directly’ (§5.2.9)
mollay, nam mollay ‘unbeknownst to others,
secretly’ (< molu- + -ay)
musim kho ‘unintentionally, innocently’
naypta ‘violently, suddenly’
tul(.)ipta ‘forcefiilly’
hampu (lo) ‘recklessly’ (§5.2.9)
makwu, mak’ ‘carelessly, at random; hard,
much’ (? derived adverb < mak-)
140 PART I
A Reference Grammar of Korean
honca ‘alone’ (also a pure noun: ~ ka)
honca se ‘alone, by oneself (Cf twul-i se
‘as a pair’)
(il)pule ‘on purpose, intentionally,
deliberately’,
cimcis ‘on purpose, intentionally,
deliberately’
puci-cwung (ey) ‘unawares’
cikcep (ulo) ‘directly; personally’
kancep ulo ‘indirectly’
sonswu ‘with one’s own hands’ (< son +
so, variant of se)
kosulan hi ‘intact’ (§5.6.1)
kaman hi ‘quietly’ (§5.6.1)
nalan hi ‘in a row’ (§5.6.1)
canttuk ‘till full, to capacity; fully, intently’
4- impressionistic adverbs of movement and appearance (phonomimes, phenomimes - §14)
+ X hi, XY hi, X-yen hi (see entry hi in Part II)
4- N 1 N 1 -i (see entry -i in Part II)
4- NjNi, in which NiNj = cosim cosim ‘cautiously’, kwuntey kwuntey ‘in (various) places’,
pangwul pangwul ‘in drops’, ...
4- derived adverbs from inflected stems (see entries -i, -wu/-o in Part II)
4- adverbative forms (see entry -key in Part II)
4- a few infinitives (such as samka ‘respectively’) and miscellaneous inflected forms
4- deictic adverbs of manner and direction (ili, kuli, celi, ... ; §5.2.4) and derivatives
(kulek celek, ... ; see Part II)
4- deictic adverbs of degree (i-taci, ku-taci, ce-taci, ... ; i man, ku man, ce man, ... ; see Part II)
4- many adjectival nouns 4- - cek with ulo: celqtay, mlswul, ...
4- many nouns with ulo: ekci, han kaci, him, Im.uy, kong, maum, macimak, pantay, yelqsim, ...
4- many nouns with the temporal-locative particle ey: icen, Ihwu, achim, nac, pam, ...
I have not listed a category of adverbs of place. Except for a few adverbs derived from adjectives,
such as melli ‘in the distance’ and kakkai ‘nearby’, and adverbs derived from iterated nouns (kos.kos-i
‘everywhere’, cipcip-i ‘in every house’) expressions of place are usually phrases of noun 4- particle -
typically ey, (ey) se, ulo, but also other particles - or place nouns used in absolute position. But this
is true also of many of the adverbs of time listed above (though not all of them); the classification
obviously needs refinement. The principal criterion to indicate an adverb rather than a pure noun is
that the word will not occur as subject (marked by i/ka) or direct object (marked by ul/lul). There are
rare exceptions, under unusual circumstances, as in Mence ka te coh.ass.ci? (Icey n’ kulen kihoy ka
olq kes kath.ci anh.a) ‘Don’t you wish you’d done it first? (It’s unlikely there’ll be another such
opportunity now.)’ (CM 2:120).
There are also deictic adverbs of direction (ili, kuli, celi, ... ) and deictic place nouns (yeki, keki,
ceki; eti); see §5.2.4. Compare, too, phrases with the postnouns kkili and kkes (Part II). A few strings
of two adverbs are spoken together as a simple phrase, and these are often treated as lexical
compounds: com-te ‘a little more’, tto-tasi ‘yet again’, tewuk-te ‘still more’, motwu-ta ‘all’, coy-ta
‘all’. CF Mkk 1960:7:34.
5.2.6. Bound adverbs (preverbs, verb prefixes).
The few morphemes that are prefixed to inflected stems are here called bound adverbs or
preverbs, but they are usually treated as verb prefixes. Eleven attach to verbs of the processive type
only, but es- and hes- function also as bound nouns with the postnominal adjective toyta:
cat- ‘small, fine’ (as adnoun, see p. 150)
cis- ‘hard, severely, roughly’
es- ‘crooked, ... ’ (also adnoun)
hes- ‘vain, mistaken; mis—■ ; open’ (also adnoun; = heq- < he < he ‘empty’)
hwi- ‘round and round; enveloping; thoroughly, completely; recklessly’
pi-, pis- (just a spelling alternant?) ‘crooked, ... ’ (also adnoun)
toy- ‘back, again; in reverse’
twi- ‘back(wards); extremely; recklessly; thoroughly, completely’
(< tw! < "twuyh ‘behind; excrement’)
A Reference Grammar of Korean
PART I 141
tes- ‘additionally’ (also adnoun; = teq- < te ‘more’)
tey- ‘incompletely, partially, unsatisfactorily’
tul- ‘hard, violently, thoroughly’
tul.i- ‘hard, extremely, recklessly, suddenly; into, inward’
Three preverbs attach to descriptive verbs (adjectives) only:
say(s)-/si(s)- ‘vivid, deep, intense’
tu- ‘very’ (emphatic)
yal- ‘despicably’
The extended forms of say(s)-/si(s)- are treated as sayq-/siq- by the North Korean spelling system.
I suspect that the final -s of es-, pis-, and perhaps cis- are also -q (as that of tes- more obviously is),
although they are written with -s in the North Korean system, too. The -s of mos ‘cannot’ could also
be from -q (Cf mo-cala- ‘be insufficient’), despite the noun calmos ‘mistake’, which is a later
formation (see §4.2, §5.2.5), and the earlier spelling "mwot. (That is, the earlier final -t in some
instances may be another form of the same marker of subordination as the -s that is ancestral to most
cases of -q.)
There are several pseudo-preverbs of transparent derivation:
yes- ‘on the sly’ (obsolete stem ‘spy on’)
chi- ‘upward’ (obsolete stem ‘ascend’)
che- ‘abundantly, thoroughly, extremely, severely, at random, without permission or cause’
( < chye = chie, vt infinitive ‘hit, ... ’)
ey- ‘surrounding’ (< eywu-, vt ‘surround’)
kala-(anc-) ‘sinking’ (a bound infinitive, perhaps < kal.e ‘plow [under]’?)
kule- ‘pulling; clutching’ (variant of kkul.e, vt infinitive)
salo ‘alive, awake’ (derived adverb < sa-1- ‘live’ + suffix -o = -wu)
elwu ‘caressing’ (derived adverb < elu- ‘pamper’ + suffix -wu), not to be confused with
the archaic adverb e 'Iwu ‘possibly’ (see Part II)
As an adverb il ‘early’ seems to be limited to the expression il kkay- ‘wake up early’ but il also occurs
as an adnoun, and ‘early’ is usually expressed by the adjective ilu- or the adverb il-ccik. Some of the
regular monosyllabic adverbs (cal ‘well’, mos ‘definitely not, cannot’, tel ‘less’, ... ) might be taken to
be preverbs, but they are saved from the tag “bound” because of their wide distribution: they freely
occur in new formations, and most of them can be separated from the verb by focus particles or the
like. Certain compound verb stems (verb + verb) might be misviewed as adverb + verb, especially
those with obscure etymologies such as these: pulu-cic-, pulu-cwi-, pulu-thu-, pulu-pttu-; ppom-nay-
(< ppop-may-); tha-ilu-; ce-peli- (< cie peli-); momc(y)e-nwuw- (mom ul cie nwuw-). CM 1:421
lists (p)pet- as the equivalent of a bound adverb which means ‘out(wards), mistakenly, mis-’ in (p)pet-
ka-, pet-na-, pet-noh-, ppet-titi-, ppet-chi-, ppet-su-; also sel- (< se-1-) with the meanings ‘half-
cooked, raw; unfamiliar’ in sel-teychi-, sel-salm-, sel-ik-, sel-talwu-, sel-mac-, sel-capcoy-.
A few other bound elements attach to the beginning of verb stems, such as alo- in alo-sayki- and
ek- in ek-nwulu- and other verbs; each poses special problems. The bound element epsin- in epsin-
yeki- ‘disdain, slight, neglect’ obviously comes from eps.i + yeki-, thus ‘treat as nonexistent’. The
Hankul spelling pins the irregularity on the first element, but historically it is properly placed with the
second, and I would prefer to write epsi(-)nyeki-. The LHS dictionary implies that the pronunciation is
/•••nny •••/; if that were true, we would have to write epsinq yeki- but no other sources confirm that
pronunciation. Some sources (Cosen-e so-sacen, Kwuk.e say sacen, ...) give the initial vowel as short.
The first element in hu-nukki- ‘sob’ seems to be a truncation of the phonomime huk (huk) ‘sob
(sob)!’. Somewhat obscure elements: to- in to-math-; tong- in tong-calu-; hol-/hwul- in hol-kapun
ha- and hol-potul / hwul-putul ha- (see entries in Part II). The cen in N ey cen kkam-kkam hata ‘is
completely ignorant of N’ is a shortening of cen-hye < cen ha.ye.
For other “prefixes” see the bound adnouns of §5.3.3.
142 PART I
A Reference Grammar of Korean
5.2.7. Interjections.
An interjection is a subclass of adverb that typically occurs by itself as a minor sentence, often
with the exclamation-point intonation (§2.8) and special voice qualifiers (not treated in this book),
sometimes with expressive length (§2.7.2) or abrupt end - the glottal catch, Romanized as final -q.
I have not made a special study of this part of the lexicon, so that the words given below (mostly
without translations) are largely taken from the works of others, notably Choy Hyenpay 1959:581-2
and Kim Pyengha 1:266 ff. Among the items are a few words that are transferred bodily from other
parts of speech: coh.ta ‘it is good’ = ‘fine!’. The interjections are arrayed in semantic categories; a
closer study would probably lead to a rearrangement. Notice that the particle una / na is often used to
emphasize adverbs and interjections; examples will be found in Part II.
(1) calling people
(2) calling animals
(3) shooing animals
(3a) shooing people
(4) yes to call
(5) YES to command
(6) YES to question 1
(7) ‘Yes of course’
(8) no to question 1
(9) ‘Not sure’
(10) hesitation
(11) urging, inviting
(12) encouraging
(13) damning
yeposipsio!, yeposey yo!, yeposio!, yepo! (in
descending order of politeness)
i p(w)a!
i ay!
ana! ‘hey!’ (S. Kyengsang dialect)
i! ‘hey look out!’
ungya! [uya] (friendly, women to women)
swi(-swi)!, swis! [swiq] ‘psst!’, ‘hush!’
kwukwu! (chickens)
weli! (dogs)
olay olay! (pigs)
ile! (< il.e) ‘giddyap!’ (to horse or ox)
neymi! ‘here calf!’ - also (13)
i kay! ‘get, dog!’
i kwayngi, i koy! ‘get, cat!’
swe!, hwei! ‘shoo, birds!’
ya ya! ‘out of the way, you kids!’
yaytula! pikhyela! ‘get out of the way!’
ney! (yey!), nay!; kulay!, ung!; way!, mue!
(in descending order of politeness)
ney! (yey!), nay!; kulay!, ung!; onya!, o!, I! [IT]
‘all right! OK!’ (in descending order of
politeness)
ney! (yey!), nay!; kulay!, ung!
am (mullon id, kuleh.kwu malkwu)!; amulyem!
ani olssita!, ani (ey) yo!, ani!
kulssey olssita!; kulssey yo!; kulssey!
cham ‘uh; oh!’
um ‘hmm’
ce, ca; ceki ‘uh’
ka se (nun / Hang / Hang un) ‘and uh’
mue (’la ko / ’la ’nun / ’la ’n’ / ’la ’lq ka) ‘uh’
mal ita, mal ia, mal ya ‘I mean, you see, you know’
ca!
wiye!
pethye la! ‘hold out!’
ppop-nay la! ‘be proud!’
i-nom!, ku-nom!, ce-nom!
neyncang!, yeyncang!, ceyncang (mac.ul/chil)!
neyki!, ceyki!, neymi! 2
A Reference Grammar of Korean
PART I 143
(14) disgust,
dissatisfaction
(15) censure
(16) sneer
(17) snicker
(18) rejection
(19) effort
(20) pain
(21) fright
(22) anger
(23) disappointment
(24) pity
(25) denial
(26) recall
(27) recognition
(28) surprise
(29) sigh
(30) laughter
(31) delight
(32) approval
(33) sarcasm
on!; ey!
chi!; si!; eysi!
atta!; aytta!; eytta! - also (16)
aykay(kay)!
eti 1’!
eytta! - also (15); eyla!
he(he)!; phi!; phu!; hwu!
pikhye la! ‘out of the way!’
ai(kwu)!; chwi!
i(y)eeha!, i(y)engcha!, iungcha!, yecha!, e.yecha!,
ekiyacha!, iyessa!, e.yessa!
ayko!, aiko!, aykwu!, aikwu!, eykwu!, ei(kwu)!, aywu!
ii!, wuwu!
eypi(ya)!, eyttukela!
ikki (na)!, ikhi (na)!
ey!, eys! [eyq], eyik!, eyk(.)ki; wen!
e(ng)!
apulssa!, eppulssa!
acha!, echa!
aacha (aikwu)!, aikya!, aykay!, eyku!
celen!; haha! - also (30)
eti!, etten!; weyn ke 1’!
chenman ey (yo)! ‘[not] in ten million = not at all’
eyla!; as.a la!
cham!; ceng-mal!
kuleh.ci!; kulem!
amulyem, am
a(a)!; ak! [aq]; ai!; yaa!
ayko!, aiko!; aykwu!, aikwu!; eykwu!, eikwu!
ema (na)!, erne na!; eykwumeni!
ikki (na)!, ikhi (na)!, ikhu!
atta!, eytta!; watta!
celen!; ke!
he(he)!, ha(ha)!, hwu!, hwuyu!
he(he)!, ha(ha)!, hi (hi)!, ho(ho)!
aa!; yaa!
eyla!; eyla coh.kwun a!; eyla manswu!
mansey!
coh.ta! - also (32)
coh.ta! - also (31)
cal hanta!
olh.ta!, olh.ci!
elssa! - also (33); elssikwu (na)!
ikhi!, ikhwu!, ikhu!
aykhay!
eyttwu eytta! (dialect?)
a(c)cwu! ‘and how!’
elssa! - also (32)
khayssamey! (Kyengsang; < hakey hay ssamyen se)
144 PART I
A Reference Grammar of Korean
elepsyo! (< elim eps.e yo)
yongyong!
allangchong!
aii!
uak! ‘boo!; puke!’
u(ng)a! ‘bawl!’ (of baby)
(kaychiney) sswey! (said after sneezing)
koswuley! (kosiley!, kkosiley!, kosiney!) 3
1 As in many languages (but not English) the reply to a negative question accords with the surface
structure (“Yes, we have no bananas”) rather than the underlying meaning, unless the negativization
is merely rhetorical, as in an invitation.
2 Some of these words have vivid etymologies: neymi comes from ne uy emi - ‘your mother ■■■ ’.
3 This is said: (1) when performing a shaman rite; (2) when eating in the open country; (3) shortly
after leaving a house from which food is taken. Kosi was the legendary teacher of farming.
To the lists can be added mimetic adverbs like kkaok ‘caw!’ and other phonomimes like swi!
swiya! ‘tinkle-tinkle!’ (sound of child urinating), and occasionally some of the phenomimes (§12).
5.2.8. Bound nouns.
We will call certain elements bound nouns and bound compound nouns (Cf §5.2.1). Of these,
many are bound adnouns (§5.3.3) and bound postnouns (§5.4.5). Others are bound as subjects or
objects (pal-petwungi in pal-petwungi chi- ‘stamp one’s foot’), and some are of doubtful classification
(cheng ‘membrane’, pho ‘quantity’ - see Part II). There are also bound adjectival nouns, or
“postadjectivals” (§5.6.5); Cf the bound postverbs (§7.1.3). The word mangceng ‘although’ occurs as
a postmodifier (§5.4.3) and also, probably as a reduction of -ki ey ’1 mangceng, after the summative
form -ki + particle ey; it, too, is a kind of bound noun. Notice also the quasi-adnouns (§5.3.1) and the
bound preparticles (§5.2.9).
Other bound elements of obscure etymology include the “pre-postnominals” alum in alum-tapta
‘is worthy of alum = is beautiful’ (perhaps from alam ‘tree ripe’ < al pam ‘shelled chestnut’) and
alis (or aliq) in alittapta ‘is worthy of ali(s) = is charming’ (< ?); the -thi in kokay-thi ‘a steep
twisting road over a mountain ridge’ (Cf Japanese [mi]-ti ‘road’ and ti-mata ‘crossroads’).
The morpheme tes ‘interval of time’ is treated as a bound noun (rather than, say, a quasi-free
noun) because it occurs only in the compounds enu-tes ‘before one knows it’ and tes-eps.ta ‘is
ephemeral’ (no particle permitted to intrude). Possibly similar is nacel ‘half-day’, which seems to be
limited to the expressions (1) han nacel ‘one/a half-day’, pan nacel ‘quarter-day’; (2) achim/cenyek
nacel ‘in the course of the morning/afternoon’; (3) nacel kawus ‘the better part of a day’.
Certain bound nouns (suffixes) are limited to one or just a few nouns, as is true of the many
vulgarizers, the constituents of mimetics, and categorizers like - cin(i) ‘falcon’. Such elements are
often of obscure etymology: can satali (= can sasel = can mal) ‘small talk’; kacis-/kecis-puleng(i),
-puli = kacis/kecis mal ‘lie, falsehood’; kho-mayngnyengi/-mayngmayngi ‘one who speaks with a
nasal twang’; pola in nwun pola ‘snowstorm’ and mul pola ‘spray of water’; sal in mom sal ‘general
fatigue’; sali = coki ‘yellow corvina’ in polum sali ‘a coki caught at midmonth (high tide)’ and
kumum sali ‘a coki caught at the end of the month’. Some nouns are probably the result of
lexicalizing a phrase: pal ssasim ‘fidgeting’ perhaps < ‘foot (deigning to) be swift’? Probably a variant
(or special use) of a Chinese verbal noun: yak sisi ‘administering medicine’. The noun kophayngi has
three meanings, and there may be more than one etymology: ‘coil; round trip’ ?< ko ‘loop’ +
-phayngi; ‘the critical moment, the climax’ = kop(-)i < kop- ‘bend, turn’ + -phayngi or -h-ayngi.
Sometimes the second element seems to be an obscure noun that is being explained by the first
element: ip-swul < ip siwul < ip si 'Gwul ‘the edges (line) of the lips’ (? < ‘bowstring’); nwunq sep
(nwun-ssep) ‘eyebrow’; sin-pyena ‘the stitched part of a shoe’; soy-sulang ‘a forked rake’; pyen-cwuk
‘rim, brim’; twi-thongswu ‘the back of the head’; thopq yang ‘saw blade’ (probably a diminutive
(34) ingratiation
(35) other
A Reference Grammar of Korean
PART I 145
*ni-ang < ni ‘tooth’); mith celmi ‘basis, foundation’; oymyen swusay ‘flattering’; ocwum sothay
‘diuresis, a weak bladder’ (Cf sothayk ‘swamp’). The expression tamq pyelak = tamq pyek ‘wall;
blockhead’ seems to contain either an expanded variant of pyek < ‘pyek ‘wall’ or a variant of pyelang
< pyelh ‘cliff’; Cf sonq pyek “hand wall” = ‘the flat of the hand’. Sometimes an obsolete noun is
found: nwun-sselmi < nwunq selmi ‘a quick eye (for learning things)’ < selmuy = sulki < sulkuy
‘sagacity, good sense’. The second element of ip sim ‘boldness/brazenness of words; eloquence’ and
payq sim ‘impudence, nerve, chutzpah’ is a dialect variant of him ‘strength, power’.
The second element of cho-sung ‘first days of the month; newborn’, i-sung ‘this world’, and ce-
sung ‘the other world’ is a variant of (sayng <) soyng ‘life, ... ’. The second element of ssi-as ‘seeds’
(not attested before 1775) is probably a lenition - Gas from ka'cfi] ‘variety’. The noun namnye-
chwuni ‘hermaphrodite’ may have a variant of -chwungi = -cwungi or (directly) of cwung <
cywung ‘middle’ + i ‘person’. The second element of pic-cisi ‘intermediary party to a loan’ looks like
a derived noun *cis.i (dialect < ci'zi; the standard version should have been *cii) from cls.ta ‘makes’.
But it might be els ‘act, gesture’ (probably the stem of cis.ta used as a noun) + i ‘person’. The noun
meyali ‘echo’ can be traced back to moy-ali < "mwoy-'zali, a compound of "mwoy ‘mountain’ +
probably either a variant of swo'li ‘sound’ or the particle s + a variant of *wu'l-i ‘sound’ (and not
sa l-i ‘living’). The ccim of mokchim ccim ‘hitting one with a wooden pillow’ and mongtwungi ccim
‘clubbing’ looks to be a variant of chim, substantive of chita ‘hits’. The -tha(y)ki of homtha(y)ki
‘crotch’ may be a variant of (t)tayki ‘stick’ attached (presumably) to horn ‘groove’. The -han(-i) of
wentwu-han(-i) ‘melon planter/farmer’ seems to be limited to that term alone, and han is usually
treated as a Chinese loan. The noun pok-cheli ‘an unlucky person’ is perhaps better pokchel-i <
pokchel ‘(repeating another’s) failure’ (< ' PWOK-'THYELQ ‘rut left by a capsized carriage’) + i
‘person’. The expression namu cicekwi ‘wood chips’ is a variant of (dialect) namu kicekwi, according
to KEd, but the composition of that is unclear; Cf cice-kkaypi ‘wood chip’, namuq kaypi ‘piece of
wood, splinter’, ccic- ‘tear, rip’. The noun humcileki ‘stringy ends of meat’ probably has a suffix, but
just where to cut is a problem: humcilek-i, humcil-eki, hum-cileki? There is no clear source for the
first part; the best candidate is hum ci- ‘get scarred, marred, flawed’ (< *"HUmI*"KEM ‘deficiency’).
The relationship of kkoli < skwo 'li and kkolang(c)i ‘tail’ with kkongci ‘tail of bird’ is unclear;
kkongci is probably a contraction of kkolangci, from kkol-ang[i] + ci (bnd n) ‘stuff, thing’. The
second element of cang-acci ‘dried radish or cucumber slices seasoned with soy sauce’ is of unknown
origin but may contain ci ‘stuff, thing’ (Cf achi, -aci, chi). Variants of chay-ccik ‘whip’ have -ccwuk
and -ccok, and the last is probably the source, perhaps identical with ccok < (p)cwok ‘piece’. How
kkun and kkunapul ‘(piece of) string’ are connected is unclear.
In Part II we treat -so of mom-so ‘in person’ and -swu of son-swu ‘with one’s own hand’ as
variants of a bound particle.
5.2.9. Bound preparticles.
The bound preparticle mo occurs in mo chelem ‘like mo = taking great pains, with much
trouble/effort; at long last’; this appears to be the only preparticle followed by chelem. The etymology
is unclear. Perhaps it is the noun mo that refers to a difficult and desirable arrangement of the four
sticks in a game of yuch; or perhaps it is somehow related to mbs ‘cannot’. But most likely it is an
abbreviation from amo = amu ‘any’; Cf mo-ccolok from amu-ccolok.
Several bound preparticles precede the particle ulo/lo and the resulting structures are treated as
unanalyzed adverbs by Korean grammarians:
sinap in sinap ulo (dialect variant sinam ulo) ‘at odd moments’ (with the earmarks of a Chinese
binom si + nap/nam but the actual etymology is obscure)
susu lo ‘of itself, spontaneously; oneself’ < susa < sa-sa < so so (‘private private’)
kakkas in kakkas ulo ‘barely’
pa.yah (note the rare final h) in pa.yah ulo ‘nearly, on the verge of; in full swing’
no-pak.i (< no(sang) ‘constant’) in no-pak.i lo (= puth-pak.i lo) ‘fixedly’
146 PART I
A Reference Grammar of Korean
Here we can include the first element of the following expressions, too:
ol[h] ulo ‘to the right’ (< adjective olh-; not *ol lo!)
oy lo ‘to the left’ (< adjective oy-)
muth ulo ‘in a lump, at one time’ (< ?; CF mus ‘all’)
thong ulo ‘all, wholly’ (CF on-thong)
ken ulo = kenseng (ulo) ‘in vain’ (etymologically identified with the adnoun ‘dry, dried’)
nal in nal lo ‘raw’ (also an adnoun, and perhaps etymologically derived from nal ‘to be born’,
the prospective modifier of na-)
hoi lo ‘alone’ (also an adnoun)
pyel lo ‘specially’ (also adjectival noun, pre-postnominal + na-)
sayng ulo ‘raw; unreasonably’ (also adverb and adnoun; a bound Chinese morpheme ‘birth’)
cel lo, ce(y)-cel lo ‘automatically, without effort’ (apparently from ce ‘oneself’)
ta-cca ko-cca lo ‘unexpectedly, without any warning, suddenly, directly’ (< ?)
maynani lo ‘empty-handed’ (? < adnoun mayn + derived adverb anh.i)
We may wonder whether to include also ka( )lo ‘horizontally’, sey( )lo ‘vertically’, and se( )lo
‘mutually’. But kalo may be a derived adverb from kalu- ‘cut across’, as palo is from palu- ‘be
straight, right’. And (k)k%kkwu(-)lo ‘upside down’ is to be connected with (k)k3akkwule (d-)
‘tumble’. (For the purely orthographic distinction of ka ulo ‘toward the edge’ from kalo, and of se ulo
‘toward the west’ from selo ‘mutually’, see the entry ulo in Part II.)
The nouns hoth ‘single(-ply)’ and mat ‘eldest’ are limited to occurrences before the particle ulo,
before the copula ita, and before the postnominal verb ci-; each is also an adnoun. There are probably
other precopular nouns, and perhaps some also occur with ulo, but they have not come to my attention.
One interesting case is tahayng, which is usual only with ulo (tahayng ulo ‘luckily’) and ita (tahayng
ita ‘is fortunate’). The word does not normally occur with other particles, but the nominative i/ka will
be present when the copula is negativized: tahayng i ani ’la pulhayng ita ‘it isn’t fortunate, it’s
unfortunate’. Tahayng is mistakenly used for yohayng ‘good luck’ in tahayng ul palanta = yohayng
ul palanta ‘gazes on (= receives) good luck’; and tahayng also occurs as an adjectival noun ‘be
fortunate’ but only in the forms tahayng hakey (to) ‘fortunately (indeed)’, tahayng hi ‘fortunately’,
and tahayng han(q II) ‘a fortunate (matter)’ - CF §5.6.1. We might wish to consider as a precopular
noun the ani of the negative copula ani ’ta ‘it is not’ (ani also occurs as an interjection); see the
discussion in §11.7.1.
5.3. Adnouns and pseudo-adnouns.
One of the environments of the noun is before another noun (or noun phrase) which it modifies.
Some words occur exclusively or typically only in that position. These we call ADNOUNS; they can also
be called prenouns. Often included by Korean grammarians are the pseudo-adnouns, some of
which are derived from reinforced forms of nouns (hays ‘new, of the year; sunny, of the sun’ < hayq
- < hay ‘year; sun’) and some of which are modifier forms of a verb that have come to emphasize
some special meaning a little more than other forms of the verb do. Korean grammarians prefer to spell
‘right (in direction)’ as olun to distinguish it from ‘right (in correctness)’, which they spell as olh.un.
I am inclined not to call most of these forms adnouns, but they are included in the lists below. In a
sense, any modifier can serve as a pseudo-adnoun, but those that are so treated by the Korean
grammarians are usually distinguished by some kind of parallelism with single morphemes elsewhere in
the vocabulary: hen ‘old’ ( < he-1- vi ‘get old, wear out’) is the antonym of say ‘new’ and a synonym
of the Chinese bound adnoun kwu- ‘old’; sen ‘half-done = immature’ ( < se-1- ‘be half-done’) is a
synonym in some contexts of the adnoun sayng, a single morpheme of Chinese origin. Certain nouns
have special, and usually shorter, shapes in adnominal position (kac’ ••• = kacwuk ‘leather’, ’mak •••
= macimak ‘last’); I have not listed these as pseudo-adnouns. Notice also the adjective construction
X-una X-un ‘that is ever so X’, limited to adnominal use (see -una in Part II).
There are three lists of adnouns: (1) those that seem to occur only as adnouns, (2) those that have
some other uses (briefly noted in parentheses), and (3) pseudo-adnouns, for which the etymological
A Reference Grammar of Korean
PART I 147
sources are indicated. There are numerous constraints on the occurrence of individual adnouns; these
have not yet been explored in any systematic fashion, but hints as to their nature can be found in both
the examples and the translations in the entries of Part II. Some adnouns, we will see, can be separated
from the following noun by other modifying phrases; others, more like the bound adnouns (or prefixes)
in this respect, cannot be separated. In the lists below, those adnouns which are clearly separable are
marked “ + ”, those clearly inseparable are marked and the intermediate or questionable cases are
marked “(+)”.
List 1: Adnouns (exclusive)
— ches ‘first’
— cey ‘-th’ ordinalizer (with Chinese numerals) (< TTYEY)
+ ku- /ko-kkacis ‘such a - as that’
+ i-/yo-kkacis ‘such a - as this’
+ ce-/ co-kkacis ‘such a - as that’
+ ney-kkacis ‘the likes of you’
cey-kkacis ‘the likes of himself/herself/themselves’ (not ‘the likes of me’!)
(+) yeys ‘old, ancient’ (< yeyq - < "nyey s; Cf noun yey < "nyey)
(+) yenu(y), yeni ‘usual; (most) other’
(+) on ‘whole, entire’ (Cf the Chinese bound adnoun cen-)
+ on kac’ ‘all’
+ * han ‘one ( *- hana); the whole; the peak, extreme, most, very; about, approximately’
*In the meanings ‘one - ; a certain ••• ’.
— han, hal ‘large, great; proper’ (modifier and prospective modifier < ha- obsolescent adjective
‘much, many’)
— han ‘outdoors, outside’
+ ku-/ko-man ‘that little ••• ’
? ku-/ko-mas ‘that little ’
— al ‘bare, ... ’
+ swun ‘pure; net’ (< ssywun)
— swus ‘pure, innocent’ (? < swufn] s)
— uypus, epus ‘step-(relative)’
— itum (nal, tal, hay) ‘the ensuing/next (day, month, year)’
(+)? kwun ‘extra, uncalled-for, excess’
— has ‘cotton-padded; with spouse’
(+) tan ‘only; single’ (but usually an adverb) (< TAN)
+ ttan ‘another, different’
+ yak ‘about, approximately’ (< 'qyak)
— yang (atul, ttal, pumo) ‘adoptive/foster (son, daughter, parents)’ (< ' yang)
(+)? kekum (+ TIME phrase) ‘ago, earlier, back (from now)’ (< 'KE-KUM)
— tol ‘wild, rough; untutored; inferior’ (? variant of tul, below)
(+)? may ‘quite, much (the same)’
(+)? may ‘each’ (< " moy )
(+)? kak‘each, every’(< 'kak)
(+)? mo ‘(a) certain’ (= amo = amu; but from mo < "mwuw)
+ oman [? dialect] ‘whole, all, every, many’ (? < on + manh-; ? < o-man ‘50,000’)
(+) mus < mwu[l] s ‘many, all sorts of’
+ musun ‘what; some one ••• ’
+ enu ‘which; any’
(+)? tong ‘the same; the said’ (< ttwong)
(+) pon ‘this; main; real’ (< "PWON)
148 PART I
A Reference Grammar of Korean
(+) hyen‘the present (existing, actual)’(< ' HHYEN )
(+) wen ‘the original’ (< ngwen)
- cay ‘resident in’ (< CCOY )
- tok ‘by oneself, alone’ (< ' ttwok )
- pay-nay(q) ‘newborn’ (< noun + vc. ‘expel [from] belly’)
- si ‘one’s husband’s (relatives)’ (< ssi)
- soy ‘a small one’ (Chinese bound adnoun so- < "sywow + particle uy)
- soy ‘of cattle’ < "sywoy < syw[o] 'oy (§2.12.3) < *'sywo oy.
- (k)kamak ‘black’ (< adjective (k)kam- + -ak)
- ong, ongtal ‘small and sunken’
- (p)palkan, (p)pelken ‘utter, downright’ (< adjective modifier)
- ppetuleng ‘out-turned’: ppetulengq (n)i = petq-(n)i = tesq (n)i ‘buck teeth’,
ppetuleng i ‘person with buck teeth’
- kalang ‘fine, tiny, shriveled’; (= kalangi) ‘forked’
- yang ‘foreign, western, Occidental-style’ (abbreviation < se.yang < sYEY- yang)
- n ye ‘woman, female’ (abbreviation < n yeca < "ATE- "CO)
- nam ‘man, male’ (abbreviation < namca < nam-"co )
- mok ‘wood(en)’; (= mok.myen) ‘cotton’ (< 'MWOK)
- ho ‘of foreign origin, especially from ancient China’ (< hhwo)
- tang ‘of Chinese origin’ (< noun ‘Tang dynasty’) (< TTANG)
- pem ‘pan-, all-’ (< ppem)
(+) 'yang ‘both’ (numeral) (< "LYANG)
- 'nayng ‘cold, iced’: 'nayng khephi ‘iced coffee’, 'nayng saita ‘chilled cider’, 'nayngq kwuk
‘cold soup’(< ’■ loyngIlyeng )
List 2: Adnouns (non-exclusive)
+ i/yo ‘this’ (noun + particles)
+ ku/ko ‘that’ (noun + particles)
+ ce/co ‘that’ (noun + particles)
(+) say ‘new’ (noun + lo, + low-)
- oy ‘only, single’ (noun + lo, + low-; bound adverb + tte-1-)
(+) amu ‘any’ (as noun ‘any person’)
- mat ‘first-born, eldest’ (noun + ulo, + ita)
(+) cen ‘former’ (also noun, postnoun; < CCYEN )
- ’mak-nay ‘last-born, youngest’ (? noun)
- cho ‘of the first ten days of the month; early’ (also postnoun; < chwo)
- ay ‘the very young; (= a.yey) the very first’ (also noun, abbreviated < ai ‘child’; bound noun
in ayq toyta)
- tul ‘wild’ (as noun ‘prairie, moor’)
- cin ‘deep (in color)’ (also adjectival noun ‘be deep or thick’; < cin)
- yen ‘light (in color)’ (also adjectival noun ‘be light or soft’; < " zywen )
kay ‘wild, ••• ’ (noun ‘dog’)
(+) ko ‘the late (deceased)’ (also noun and postmodifier ‘reason’; < 'Kwo)
- phus ‘green, unripe, ... ’ (also bound noun in elyem-phus; ?< phuflj s)
- 01, o’ ‘early-ripening’ (also noun ‘vigor, ... ’)
- il ‘early’ (also adverb; Cf ilu- ‘be early’, ilccik < ilq-cik)
- nuc ‘late, belated’ (< adjective nuc-; also adverb?)
- tes ‘additional’ (also bound adverb) (< teq ••• < te ‘more’)
- cap ‘mixed; poor; ... ’ (also bound noun + toy-; < ccap)
A Reference Grammar of Korean
PART I 149
- pi(s) ‘crooked, ... ’ (also bound adverb)
- es ‘crooked, ... ’ (also bound adverb)
- mey ‘nonglutinous’ (also bound noun in mey ci- ‘be nonglutinous’, mey-malu- ‘be fallow’)
- cha(l) ‘glutinous’ (also bound noun in cha’ ci- ‘be glutinous’; < cha- ‘sticky’)
- thong ‘whole, intact, untouched’ (also adverb, bound noun)
- hoth ‘single’ (also noun -I- ulo)
- mac’ ‘facing, ... ’ (also adverb; abbreviation of derived adverb macwu < mac-wu)
- sang ‘common, ordinary, ... ’ (also bound noun; < ssyang)
+ yele ‘numerous’ (= numeral yeles)
- pan ‘half’ as in pan pengeli ‘half-mute’ (also numeral ‘half’, postnoun ‘and a half’; < 'pan)
- nal ‘raw’ (also noun + lo; ? < nalq prospective modifier < na- ‘be born’)
(+)? sayng ‘raw; crude; live; real; arbitrary; utter’ (also noun + ulo, adverb + mek-; < soyng)
- emci ‘main, principal’ (also noun = emci kalak ‘thumb’)
- V 611 (prenumeral) ‘continuing through, running’ (also noun ‘continuation’, vnt ‘connect,
continue’; < LYEN)
- kang ‘forced; unadulterated, plain; dry’ (also bound adverb + malu-; < 'kkang)
- ken ‘dry, dried’ (also bound preparticle, adjectival noun; < kken)
- hes ‘false’ (also bound noun; < heq ... < HE s)
- swu(h) ‘male’ (also bound noun ‘convex, external, protruding’)
- am(h) ‘female’ (also bound noun ‘concave, internal’)
- kayk ‘uncalled-for’ (also bound noun; noun ‘guest’ < 'khoyk)
(+)? kun ‘about, nearly’ (also bound noun?; < "kkun)
- chain ‘real, true, genuine’ (also adverb, interjection, ? noun)
- pyel ‘special’ (also bound noun; < ‘(P)PYELQ)
+ pulkwa ‘only, merely’ (+ quantity) (also adjectival noun; < 'PWULQ-'KWA)
- sen ‘first, prior’ (also noun ‘first move’; < SYEN)
? seng ‘Saint’ (also noun; < 'SYENG)
(+) chin ‘sharing blood ties’: chin apeci ‘blood father’ (also adjectival noun ‘intimate’; < CHIN)
- oy ‘maternal’: oy hal-’meni ‘maternal grandmother’ (also bound adnoun ‘external’, noun =
pakk ‘outside, ... ’; < 'ngwoy)
(+) tang ‘the said; the appropriate; (age) at a time’ (also postnoun ‘for each’, vnt ‘undergoes,
confront; copes’, vni ‘confronts’, adjectival noun ‘is reasonable, appropriate’; < tang)
List 3 : Pseudo-adnouns
(+) taum-taum ‘next but one’ (noun + noun)
+ pyel-pyel, pyel-uy pyel ‘special’ (bound noun -I- ...)
- cey-il ‘first, prime’, cey-I ‘second’, ... (see Numbers, §5.5)
+ nay ‘my’ (abbreviation < na uy)
+ ney ‘your’ (abbreviation < ne uy)
+ cey ‘one’s own’ (abbreviation < ce uy)
? i-nay ‘this my ; my’
- hay, hays (< hayq < hoy s) ‘new, of the year; sunny, of the sun’
- wi(q), wis ‘upper’ (< noun wi ‘above’). The variant wu(q), found in only a few set phrases
nowadays, is treated as an adnoun wus in both North and South Korea. The NK dictionary
treats wi as dialect, wu as standard. Cf Mkk 8:42 (1960): “wi is used as the noun in the
central area but wus/wis when it is the prefix [or adnoun]”. The NK dictionary standardizes
the spelling wus (as do the South Koreans) where one would expect the apostrophe to write
the -q, and that spelling indeed turns up in CM 1:226, where wuq meli (“wus-me-li”) ‘upper
150
PARTI
A Reference Grammar of Korean
head’ is cited along with alayq meli ‘lower head’, which is spelled with the apostrophe in
NKd, as well. The Middle Korean form was wufh] s.
- alay(q), alays ‘lower’
- phalang / pheleng ‘blue’ (noun)
(+)? (k)kemceng ‘black’ (noun)
- nolang / nwuleng ‘yellow’ (noun)
- ppalkang / ppelkeng ‘red’ (noun)
- com ‘petty’ (noun, abbreviation < cokum)
? taymo han [rare] ‘important, main’ (defective adjectival noun)
? aymen [uncommon] ‘off-the-point, extraneous, devious (remarks), vague’ (abbreviation <
aymay hen = aymay han ‘ ■■■ that is vague’
(+) eps.nun ‘impoverished, needy’ (< ‘ ••• that lacks’)
+ hanta ’nun, hanta ko (ha)nun ‘admitted to be capable (strong)’
- sen ‘half-done = immature’ (modifier < se-1- vi)
+ hen ‘old; worn out’ (modifier < he-1- vi)
- palun ‘right’ (modifier < palu- adjective)
- olun ‘right’ (variant of olh.un, modifier < olh- adjective)
- oyn ‘left’ (modifier < oy- adjective)
(+)? can ‘small, fine, thin’ (modifier < ca-1- adjective)
(+) talun ‘other’ (modifier < talu- adjective)
+ kac.un ‘all’ (modifier < kac- = kaci- ‘hold’)
+ motun ‘all’ (< mot.un = moin ‘gathered’)
- ongkun ‘whole; intact, original; untouched’ (modifier < onku-l- adjective)
(+)? mayn/mln [also with short vowel] ‘nothing but, unadulterated, bare’ (modifier < ml-)
- mayn ‘all the way, the very, the extreme’ (? from preceding entry)
(+) kin-kin ‘long long ’ (iterated modifier < ki-1- adjective)
(+)? mak-talun ‘dead-end’ (modifier < mak-talu- ‘come to an impasse’, compound vi)
hethun ‘silly’ (< heth.un modifier < heth- = huth- ‘scatter’)
ol ‘this, the present; next, the coming’ (prospective modifier < o- ‘come’)
- wang ‘big, king-size’ (noun ‘king’; < ngwang)
- mal ‘big, large-size (animal or plant)’ (< noun ‘horse’)
? mu’ ‘light, watery’ (< noun mul ‘water’; Cf mulk- ‘thin, watery’)
Certain morphemes that might be thought to fall in the category of adnouns we will treat as free
nouns: cheng ‘green or blue’, hong ‘red’, huk ‘black’, payk ‘white’; ceng ‘real, true; center, ... ’; ... .
See also omphak/wumphek in Part II.
CM 1:212 lists cas ‘small, fine’ with two examples: cas cwulum ‘crease’ and cas cing ‘small
shoe-nail’. South Korean dictionaries treat this element as cat-, a variant from ca-1- ‘be fine’. It could
also be viewed as a shortening of cati can ‘ ••• that is quite fine indeed’; as < calq ‘to be fine’; or <
caq (= ca-1-). The spelling with t is supported by cat-talah.ta ‘is rather fine’.
The words kas ‘just (+ ages by ten)’ and tan ‘only’ are usually adverbs; see §5.2.5.
5.3.1. Quasi-adnouns.
Some of the Chinese nouns which have the earmarks of freedom, being of two or more syllables,
are nonetheless restricted to positions modifying a noun or noun phrase, and occasional examples in
other positions are to be dismissed as awkward at best. Among these “quasi-adnouns” are the defective
verbal nouns (§5.6.2) that have only the modifier form: taymo han ‘prominent’, momo han
‘celebrated’.For ‘unique’ we find both yuil han and yuil uy (as in ~ mokcek ‘unique goal’); also
yuil mui han (less commonly yuil mui uy) ‘unique and unmatched’.
A Reference Grammar of Korean
PART I 151
Other quasi-adnouns are listed below in several groups:
(1) Quasi-adnouns with uy
pulhwu uy ‘undying, immortal’
pulphay uy ‘unvanquished, unconquerable’
cisang uy ‘supreme, sublime’
celqsey uy ‘peerless’
micung-yu uy ‘unprecedented’
pul.yo pulkwul uy (? pul.yo uy,
? pulkwul uy) ‘indomitable’
pulka-pun uy ‘indivisible’
(2) Quasi-adnouns without uy
kwukcey (muncey) ‘international (problems)’ kanay (kongep) ‘domestic (industry)’
wensi (sahoy) ‘primitive (society)’ "yelyu (cak.ka) ‘woman (writer)’
coki (chilyo) ‘early (treatment)’ ilqtay (cangkwan) ‘grand (sight)’
(3) The bound postnoun cek ‘-ic’ produces compounds which are somewhat like quasi-adnouns that
do not take uy, but they can also occur:
before ulo, as in kwahak-cek ulo ‘scientifically’;
before forms of the copula especially the modifier, as in kwahak-cek (in) thayto ‘a scientific
attitude’ and thayto ka kwahak-cek ita ‘has a scientific attitude’;
before the nominative marker i/ka when followed by a negativized copula expression: thayto ka
kwahak-cek i ani ’ta ‘does not have a scientific attitude’.
(4) The bound postnoun sang ‘—wise, with respect to’ (the basic vowel length of the morpheme is
usually suppressed when it functions this way) creates compounds that are often used like adnouns, but
they may be separated from the modified noun by the copula modifier in or (more commonly) by the
particle uy, as in I ke n’ kyengcey-sang (uy) muncey ’la ko polq swu iss.ta ‘This can be looked at as
a question relevant to economics’. (Contrast the behavior of cek ‘-ic’, which never occurs with uy.)
The compounds also occur before various particles, but usually the particles are present because they
are required by later elements of the sentence. The sang compounds, unlike the cek compounds, do not
occur as predicate complement with - ita ‘it is - ’, nor with most conversions of the copula other than
the adnominal in (and substitution by uy is more common) and such adverbial conversions as ina and
the negative ani ’la. See the entry in Part II for examples.
Why are sang and cek treated as “bound” postnouns, rather than free? Mainly because in general
they do not attach to non-Chinese elements, though there are numerous contrary examples, and sang
may be more versatile than I have allowed, as is its Japanese counterpart (•- zyoo). The compounds
that result from attaching the bound postnoun are special kinds of quasi-adnoun.
5.3.2. Numerals. See §5.5.1.
5.3.3. Bound adnouns (prefixes).
The occurrence of specific adnouns is restricted in various ways. In general, I have treated as free
all those adnouns that are not of Chinese origin and in addition those Chinese adnouns which are
widely used to modify nouns of non-Chinese origin as well as those that are Chinese. Some of the free
adnouns are restricted to a rather small group of partners they can modify, others are quite productive.
I have set up only one non-Chinese bound adnoun: yel- ‘young, new’ (of unclear etymology, see the
entry in Part II). But there are a lot of Chinese bound adnouns and they are quite productive, though
they combine almost exclusively with Chinese vocabulary. When one of them occurs widely also with
non-Chinese vocabulary it is included in the list of free adnouns (for example sayng). But certain fairly
free nouns, such as cen ‘before’, hwu ‘after’, and cwung ‘midst, middle’ will be found included, for in
putting the list together I have been more concerned with convenience than with consistency.
152 PART I
A Reference Grammar of Korean
The list of Chinese bound adnouns is arranged alphabetically, in order to display homonyms;
examples are given to illustrate just why each morpheme deserves the treatment as an adnoun. The
examples are all of occurrences with free two-syllable Chinese nouns; when the same morpheme is
followed by a bound morpheme I do not treat it as an adnoun. Accordingly sin- ‘new’ is an adnoun in
sin-seykyey ‘new world’ because seykyey ‘world’ is a free noun, but not in sinmun ‘newspaper’
because mun is not a free noun (at least in this meaning). When a one-syllable bound adnoun is
combined with a one-syllable free noun of the Chinese vocabulary, it is difficult to decide whether to
treat the string as one word or two, for often the two-syllable string is more common and older than
the free occurrence of the noun, which is sometimes based on a special meaning or a shortening of a
longer equivalent. When in doubt we can always use a hyphen. In fact, I prefer a hyphen for all cases of
bound adnouns, atleast within texts: sin-seykyey despite say seykyey which means virtually the same
thing ‘a new world’. The problem of freedom or bondage of Chinese morphemes is quite vexing and
requires further study. (See the discussion in §5.4.5.) The bound adnoun pi- ‘un-, non-’, for example,
is largely limited to Chinese nouns, but it can be found for a few recent loanwords of English origin,
such as pi-Kaythollik(-)kyey se nun ‘in non-Catholic circles’, where the bondage of -kyey ‘world’ is
also in question. Similar problems of free versus bound occur in cho-inkan(-)cek ‘superhuman’ (Cf
§5.3.1), cho-inkan(-)hwa ‘superhumanization’, and perhaps pi-yeyswulq-cek ‘unesthetic’.
List of bound adnouns
SHAPE
CHARACTER
MEANING EXAMPLES
ak-
m
bad
~ sencen ‘bad propaganda’, ~ insang ‘bad impression’,
~ phyengqka ‘bad evaluation’
cak-
ft
preceding;
yesterday
~ hoykyey-nyen ‘the last fiscal year’, ~ swuip-sey
‘the last income tax’, — sipsam-il ‘yesterday the 13th’
cay-
ft
again, re¬
~ chwulpal ‘restart’, ~ ipkwuk ‘reentry (into a country)’,
~ hwal.yak ‘reactive, active again’
cang-
&
long
~ keli ‘long distance’, ~ hayng.lyel ‘long parade
(procession)’, — sikan ‘long time’
ce-
short; low
~ sokto ‘low speed’, ~ kiap ‘low (air) pressure’,
~ cwupha ‘low frequency’, ~ hyel.ap ‘low blood
pressure’
cen(-)j
the former,
ex-
~ puin/manwula ‘ex-wife’, ~ swusang ‘ex-premier’,
~ naykak ‘former cabinet’, ~ namphyen ‘former
husband’, ~ su’ nim ‘a former priest’
cen(-) 2
±
the entire
(Cf on)
~ (ca.yu) seykyey ‘the entire free world’, ~ (Mikwuk)
inmin ‘the entire (American) people’; ~ Sewulq-Cangan
(ul cenmang hanta) ‘(has a panoramic view of) the whole
city of Seoul’
ceng-
IE
regular, full
~ kyoswu ‘full professor’, ~ kyowen ‘regular teacher’,
~ hoywen ‘regular member’
cey-
fg
various,
several,
the [plural]
~ palmyeng ‘various inventions’, ~ mincok ‘the (several)
nationalities’, ~ muncey ‘(the) various problems’,
~ pangmyen ‘(the) several directions’
cey-
it
imperial;
imperialist
~ cengpu ‘(the) imperial government’, ~ cengchayk
‘imperal(ist) policy’
cha-
Jit
this
~ sasil ‘this fact’, ~ sahoy ‘this society’, ~ seykyey ‘this
world’
chin-
m
1 blood-related
2 pro-
— hyengcey ‘blood brother’, — pumo ‘the genetic parents’
~ cengpu ‘pro-government’, ~ Yengkwuk ‘pro-Britain’,
— Puk-Han ‘pro-North Korea’
A Reference Grammar of Korean
PART I 153
cho-
m
super, ultra
— sokto ‘superspeed’, — ca.yen ‘supernatural’,
— umpha ‘ultrasonic(s), supersonic’, — inkan
‘super(hu)man’ (Cf choin ‘superman’)
chong-
m
overall,
general,
total
~ kongkyek ‘general offensive’, — sayk.in
‘general index’, — sacik ‘mass resignation’,
— tongwen ‘general mobilization’, — sanchwul ‘mass
production’; ~ maliq swu ‘total number of animals’, —
Mikwuk (uy) huk.in inkwu ‘the total black population
of the United States’
choy-
m.
most,
extreme(ly)
— wutung ‘most excellent’, — sinsik ‘ultra-modern’,
~ chemtan ‘spearhead’
ci-
£
branch
— kongcang ‘branch factory’, — sen.lo ‘branch rail line’
cik-
E
direct
— swuchwul ‘direct exportation’, — kopayk ‘true
confession’, - hyensang ‘true circumstance’; — kwuk
‘undiluted liquor, sauce, ... ’, — kkwul ‘pure honey’
con-
the honored;
your
— hoysa ‘your firm’, — philqcek ‘your handwriting’,
~ ceyan ‘your suggestion’
cong-
*1
final, last
- 'yelcha ‘the last train’, — cencha ‘the last streetcar
(or train)’, — (p)pesu ‘the last bus’
cwu-
main,
principal
~ sanmul ‘the main crop/product’, — san.ep ‘the
principal industry’, — peril.in ‘the chief culprit’,
— sengpun ‘the main ingredient’, ~ umco ‘leitmotif’
cwun-
quasi-,
acting
~ hoywen ‘associate member’, — sawen ‘junior employee’,
— kyowen ‘teaching assistant’
cwung(-)
*
middle
~ kiep ‘medium(-size) enterprise’, — hak.kyo
middle school’, — seyki ‘medieval centuries’
cwung-
heavy
— kongep ‘heavy industry’, — kumsok ‘heavy metal’,
— kikwan-chong ‘heavy machine-gun’, ~ 'notong
‘heavy labor’
e- *
m
the esteemed;
your
*
— kaceng ‘your home’, — kwuk.ka ‘your nation’,
— ceyan ‘your suggestion/proposal’, ~ puin your wife’
Japanese usage, sarcastic in Korean.
ha(-)
T
bottom, lower;
last, later, ...
See Part II, p. 514.
hay-
the said
— hak.kyo ‘the said school’, — saqken ‘the incident in
question’, ~ inmul the said person’
he-
J®
false; sham
(Cf hes)
— phungsel ‘false gossip’, — yengsang = hesang ‘virtual
image (in optics)’
ho-
good
— hyelqsayk ‘good complexion’, — inmul ‘good person’,
~ insang ‘good impression’, — kihoy ‘good opportunity’
hwal-
s
&
living, live
— hwasan ‘live volcano’, — mutay ‘legitimate stage’
— sintek ‘active grace’
hwu(-)
the later
— panki ‘second term’, — seyki ‘later centuries’,
— hayng.lyel ‘after-column’
T-
*
different
~ punca ‘foreign element’, — incong ‘different (alien)
race’, — pun.ya ‘different field’, ~ mincok ‘alien race’
>in-
neighboring,
nearby
— chonka ‘neighboring cottage’, — pulak ‘neighbor
community’, — wupang ‘nearby friendly nation’
154 PART I
A Reference Grammar of Korean
ka-
m
1 false,
pretend,
fake
2 makeshift,
temporary,
provisional
~ hyengsa ‘fake detective’, ~ munse ‘false document’,
~ cwuso ‘false address’, ~ uysa ‘quack doctor’,
~ sacang ‘phony company-head’
~ kenmul ‘temporary building’, ~ kyosa ‘temporary
instructor’, ~ tolo ‘temporary road’, ~ sisel
‘makeshift facilities’, ~ cengpu ‘interim government’,
~ ip.hak ‘provisional admission (to a school)’
kang-
hard, tough
~ oykyo ‘firm diplomacy’, ~ taychayk ‘strong policy’,
~ cengpu ‘strong government’
ko-
high, tall
~ cwupha ‘high frequency (wave)’, ~ hyel.ap ‘high
blood pressure’, ~ kiap ‘high (air) pressure’,
~ sokto ‘high speed’
kong-
official;
public
~ saynghwal ‘public life’, ~ maymay ‘public
transaction’, ~ hoytang ‘public meeting place’
kwi-
1 the worthy;
your
2 valuable,
precious
~ poko ‘your report’, ~ puin ‘noble lady’, ~ kwuk.ka
‘your nation’, ~ hoysa ‘your firm’, ~ chwulphan-sa
‘your publishing house’
~ kumsok ‘valuable minerals’, ~ tongca ‘one’s precious
son’
kwii-
m
old
~ sahoy ‘old society’, ~ sitay ‘old times’, ~ seykyey
‘the old world’, ~ ceyto ‘the old system’, ~ hwaph y ey
‘the old currency’
kyeng-
m
light(weight)
~ kikwan-chong ‘light machine-gun’, ~ kongep ‘light
industry’, ~ kumsok ‘light metals’, ~ mucang ‘light
armaments’, ~ pern.in ‘minor offender’, ~ phok.kyek
‘light bombardment’
kum-
-7
the present;
this month’s
~ cwumal ‘this weekend’, ~ hayngsa ‘this event’,
~ seyki ‘the present century’
kup-
abrupt;
express
~ cenhwan ‘sudden turn’, ~ yongmu ‘urgent business’,
~ clnpo ‘rapid progress’
mal-
end, last
~ hak.ki ‘the final term (of three school terms), the
last trimester’, ~ kwicel ‘the last verse’, ~ seytay
‘the last generation’, ~ 'yelcha ‘the last train’
man-
fully, a full
~ cangsik ‘full decoration’, ~ kihan ‘full time limit,
full term’, ~ o-nyen ‘a full five years’
mang-
t
the late,
deceased
~ kyocang ‘the late principal’, ~ puin ‘one’s late wife’,
~ swukpu ‘one’s late uncle’
mi-
not yet,
un-, in¬
~ hwun.lyeng ‘untrained’, ~ kyelqsan ‘unsettled
(accounts)’, ~ kyoyuk ‘uneducated’, ~ punmyeng
‘indistinct, unclear’, ~ wanseng ‘incomplete’
mol-
&
%
devoid of,
-less,
eliminating
~ chwimi ‘tastelessness’, ~ inceng ‘inhumanity’,
~ sangsik ‘devoid of common sense’
myeng-j
noted, famed
~ paywu ‘eminent actor’, ~ sosel ‘well-known novel’,
~ thamceng ‘famous detective’
myeng -2
m
the coming;
next year’s
~ hoykyey-nyen ‘the coming fiscal year’, ~ sayngil
‘one’s next birthday’, ~ Vu’-wel ‘June of next year’
mu-
M
lacking, -less,
without
~ cengpo ‘without information’, ~ kwankyey
‘irrelevance’, ~ uymi ‘meaningless(ness)’
A Reference Grammar of Korean
PART I 155
nan-
m
difficult
— ceymok ‘hard topic’, — saep ‘difficult business’,
~ muncey ‘tough problem’
‘nay-
the coming;
next year’s
— cwumal ‘next weekend’, — hak.ki ‘next term’,
~ hayngsa ‘the coming event, coming events’,
~ sengthan-cel ‘the coming Christmas’
nay-
ft
1 internal;
secret
2 female
~ chwulhyel ‘internal hemorrhage’, ~ kwungceng
‘inner palace’, ~ punpi ‘internal secretion’
— cwucang ‘petticoat government’
oy- *
ft
external
~ chwulhyel ‘external bleeding’, ~ Mongko ‘Outer
Mongolia’, — punpi ‘external secretion’, — swuwi
‘outer guard’, — yuseng ‘outer planets’
* As free adnoun ‘maternal’; as noun = pakk ‘outside, ... ’.
pan-
jg
anti-;
counter-
~ cak.yong ‘reaction’, — hyek.myeng ‘counter¬
revolution’, — kwahak-cek ‘anti-scientific’,
— S(s)olyen ‘anti-Soviet’
pay-
w
anti-
— cengpu ‘anti-government’, ~ Upon ‘anti-
Japanese)’, — Yengkwuk ‘anti-Britain, anti-British’
phi-
w.
that
— kyengkwan ‘that policeman’, — kyosa ‘that teacher’,
— sawen ‘that employee’
phl-
WL
suffering,
undergoing
— ap.pak ‘oppression, suffering’, — senke ‘undergoing
an election’
phyeng-
¥
ordinary
~ hoywen ‘ordinary member’, — sawen ‘ordinary
employee’, — slmin ‘ordinary citizen’
ph y ey-
unworthy;
my
— hak.kyo ‘my school’, — kaceng ‘my home’,
— kwuk.ka ‘my country’
pi-
not (being);
non-, un-
~ centhwu-wen ‘non-combatant’, ~ hoywen ‘non-member’,
— hyensil ‘unreality’, ~ kongsik ‘unofficial’,
~ mayphum ‘an article not for sale’
pu-
S'J
1 assistant,
vice-
2 side, by-,
subsidiary
~ hoycang ‘vice-chairman’, — kyocang ‘assistant
principal’, — putay-cang ‘assistant commander’
~ cak.yong ‘side effect’, ~ sanmul ‘by-product’
pul- / pu-
*
not, un¬
PUL-: ~ chincel ‘unkind(ness)’, ~ hwal.yak
inactive(ness)’, — phyengtung ‘inequality’
PU-: ~ ca.yen ‘unnatural(ness)’, ~ ca.yu ‘discomfort’,
— cektang ‘unsuitable(ness)’, ~ totek ‘lack of virtue’;
Cf putong-san ‘real estate’
sa-
IA
private,
personal
~ 1 Iik ‘private interest’, ~ saynga ‘bastard’,
— saynghwal ‘private life’
sang-
_h
first of 2 or
3; earlier
See Part II.
sin-
SfT
new
— kilok ‘new record’, — palmyeng ‘new invention’,
— seykyey ‘the new world’
so-i
'J'
small, little
— cicwu ‘small landowner’, — kwuk.ka ‘small
nation’, — kyumo ‘small scale’
so-2
few, scanty
~ inq-swu ‘small number of people’, — pyengqswu
‘small number of soldiers’, — somay-cem ‘few
retail stores’
A Reference Grammar of Korean
156 PART I
ta- ^
tan- $5
tay-i Jz
tay -2 W
tha- ftfe
uy-i M.
uy -2 IS
yen-
Veng- T3
5.4. Postnouns.
A postnoun occurs exclusively or typically after a noun; in our analysis the noun modifies the
postnoun, which functions as the head of the phrase. In a sense, the postnoun is a further restricted
type of quasi-free noun (§5.2.1). The quasi-free noun hay ‘possessed thing, one’s’ would be
considered a postnoun if it were not for the fact that it occurs after nay ‘my’ rather than na ‘me’.
Some of the postnouns are taken from inflected forms; a number are also used as other parts of speech,
and those are separately listed below. Some of the items given are much more limited in occurrence
than others and they would perhaps be better listed as bound postnouns; it is hard to draw the line.
Good cases could be made for including here the following items, treated as Chinese suffixes in
§5.4.5.2: -ce ‘authored by’, -ci (periodical titles), -cok ‘tribe’, -hwa ‘flower’, -kyey ‘world’ (see
§5.3.3), -kyo ‘religion’, -phyen ‘compiled by’. Compare these with hayng ‘bound for’, which is
included here. The word ccay could be set up as a separate subcategory “postnumeral” since it nearly
always follows a numeral, just as its Chinese counterpart cey- could be set up as a separate category
“bound prenumeral”; but ccay also occurs with the adnouns ches ‘first’ and mal ‘last’ (mal ccay is
entered in LHS 920a), and it can occur after a number phrase as well as after a numeral: sey pen ccay
(or sey ccayq pen) ‘third time’ (in a more literary form sam-hoy ccay); ttek ul ney kay ccay
mek.nunta ‘I’m eating my fourth rice cake’, Cf ney pen ccay ttek i cey-il khuta ‘the fourth rice cake
is the biggest’. For the postcounter (~q) swu ‘number of ■■■ ’, see §5.5.1.
(1) Postnouns (exclusive)
achij ‘person’
achi 2 /echi ‘worth’
awus see (k)awus
ca(y)ngi ‘doer, -er, one, ... ’
ccali ‘worth; amount; a person wearing’
ccay ‘-th’ (ordinalizer of non-Chinese numerals and number phrases)
ceng kkey ‘around, about (a certain time)’
many
short
great, big,
major
against,
toward,
versus
other
pseudo-,
false
adopted,
foster
soft
the esteemed;
your
~ chwlmi ‘many hobbies’, ~ hayngsa ‘many
activities’, ~ pangmyen ‘many quarters; versatile’,
~ umcel ‘polysyllable’
~ keli ‘short distance’, ~ siil ‘a short length of
time’, ~ swumyeng ‘a short span of time’
~ cengke-cang ‘major rail station(s)’, ~ centhwu
‘major battle’, ~ hayngsa ‘big event’, ~ hwal.yak
‘great activity’, ~ kyengki-cang ‘large stadium’
~ Cwungkwuk ‘against/toward China’, ~ Mikwuk
‘toward America’, ~ oykwuk ‘toward foreign
countries’
~ cwuso ‘other address’, ~ panghyang ‘(the) other
direction’, ~ panmyen ‘the other half side’
~ sengtay ‘false vocal cords’, ~ yangphi-d ‘false
parchment’
~ camay ‘foster sister’, ~ hyengcey ‘foster brother’,
~ pep.lyul ‘adopted law’, ~ pumo ‘foster parents’
~ oykyo ‘soft diplomacy’, ~ phipu ‘soft skin’,
~ taychayk ‘soft policy’, ~ tokse ‘light reading’
~ hyengcey ‘your (or his) brother’, ~ kacok ‘your
family’, ~ puin ‘your (or his) wife’, ~ swukpu
‘your uncle’
A Reference Grammar of Korean
PART I 157
chii ‘a fixed quantity’
chi 2 ‘a general sense, a feel’
chwungi / chongi ‘one, person, thing’
ciki ‘a guard, a keeper’
cil ‘act, behavior’ (Cf noun els ‘gesture’, cl(s)- ‘make, do’)
( -q) cwung(-payki) vulgarizer (spelled -ccwung- in South Korea)
echi / achi SEE achi / echi
hayng ‘bound for, dispatched to’ [semi-literary]
he ‘approximately (a certain quantity)’ [semi-literary]
kal ‘discrimination, division, branch, kind’; [neologism] branch of study, -ology’
kalyang ‘approximately (so many), about’ (follows number)
kamali ‘a person who is the butt of ’
kan ‘an interval of ••• ; between, among’
(k)awus ‘and a half’
(k)kayngi diminutive
kkal vulgarizer (? < ~q kal)
kkes ‘to the full extent of’
kkey ‘(= ceng kkey) around, about (a certain time); near (a place)’
kkili ‘separate group (of people)’
kkol ‘at the rate of, - each, per unit’
kkwulek = -q kwulek ‘the act of’
kkwuleki ‘an overindulger in’ (? < ~q kwulek -i)
kkwun ‘a man occupied with or noted for’ (= -q kwun; *Yi Ungpayk prefers -kkwun)
koc ‘the Cape of ••• ’
kwuni ‘person’ (? < -(q) kwun + i): is pallok ~ ‘idler’ the only example?
kyeng ‘around, about (a certain time)’
may ‘shape, form, cast’
nay ‘throughout, all through (a period of time)’
ney ‘group (of people)’
ong ‘the Venerable Mister ’
pachi ‘a person with a vocation (dating from feudal days) that deals with ’
( -q) pal ‘line, streaks, rays; impression’
( -q) palam ‘without one’s •- on’ (see also below)
panciki ‘adulterated with’
pang ‘in care of’
p(h)a(y)ki, p(h)e(y)ki ‘person, thing, one; child’
pha(y)ngi ‘person, thing, one; child’ (see also payngi below)
phok ‘of the same age group; approximately, about’
ppak vulgarizer
ppel ‘the kin-relationship (standing) of - ’
sang ‘Mr, Mrs, Miss, Ms’ (suffixed to Japanese names; < Japanese san)
soswu ‘plus some (extra), ■■■ odd, a bit over ■••’(< soswu ‘a small number, minority’)
ssi ‘clan; Mr’
ta(k)ci ‘(this/that) extent, degree’
tepeki ‘lots/heaps of’
theym (them, theymi, thek) ‘as much as, all of’
thi ‘the mark (looks, air, appearance, manner) of’ (? < noun ‘dirt’; ? Seoul dialect
variant < thay ‘appearance’)
thwungi / thongi ‘one, person, thing’
thwuse(y)ngi ‘covered or smea red with’
158 PART I
A Reference Grammar of Korean
(t)tam ‘latent power; wallop’
ttawi ‘and the like, of the sort’ (< ~q tawi, der noun < taw- ‘be worthy of the name of’)
ttayki, ttuki, te(y)ki, tek.kwungi ‘one, person, thing’ [vulgar]
tteli ‘thing’ [vulgar] - Does this occur only in tung tteli ‘back’?
tung ‘and so on; the above several’ [semi-literary]
tungci ‘and such places; and vicinity’ (also quasi-free noun?)
yang ‘Miss’
(2) Rhythmically misanalyzed compound nouns
nwun-i ‘a person with eyes such that - ’ phal-i ‘a person with arms such that ••• ’
pal-i ‘a person with feet such that ••• ’ son-i ‘a person with hands such that - ’
(3) Derived nouns (§9.6) that are used only as postnouns
alh.i ‘ache, illness’
cap.i ‘taking’
kal.i ‘changing; remodeling’
keli ‘at intervals of, skipping, jumping’
kel.i ‘gait’
kel.i ‘a hanger, a stand (for)’
ket.i /keci/ ‘collection, gathering up, harvest of’
nah.i ‘a weave, a yarn of’
pak.i ‘imprinted with, ... ’
pat.i /pad/ ‘receptacle for, ... ’
phal.i ‘selling’
ppop.i ‘an extractor (pull, pincer, claw) for’
puli ‘one who works (something); work, doing, act, trick’
puth.i /puchi/ ‘of the class of, made of’
sal.i ‘living, life; garb, clothes’
ssi < (p)ssi ‘the use (state, condition, quality, mode) of’
tot.i /tod/ ‘rising’
ttut.i /ttuci/ ‘thing stripped of... ’ - Are ppye ~ and al ~ the only examples?
See also (cim) sil.i, (sil-kwup) tal.i, (slpiq) cwup.i.
(4) Inflected forms (occurring also in other environments)
chiki ‘game of hitting ••• ’ (kong chiki ‘ball hitting’)
mayc.ki ‘concluding it’ (kkuth mayc.ki ‘final touches’)
naki ‘being born in; a person from’
nayki ‘product; display; person displaying (= naki); a person from’
pat.ki ‘receiving; receiver of’
sswuki ‘a boiled dish of’ (wen-pap sswuki ‘soup with rice and rice cakes in it’)
ttaym ‘warding off; mending’
See also -nam(.)un (pp. 164, 174, 704).
(5) Used also as free nouns, sometimes with a different meaning
cen ‘before; Dear - [in letter]’ (as free noun ‘earlier time’, as adnoun ‘former’)
chey ‘style (of writing); body’ (as noun, literary)
cwuuy ‘ism, doctrine’
cwuuy-ca ‘-ist, ideologist, advocate of’
-q keli ‘material, stuff for; basis; doing; appearance’
meli vulgarizes nouns (as free noun ‘head’)
nim ‘esteemed person’ (as obsolete free noun ‘you; lover’)
nolus ‘job, role’
A Reference Grammar of Korean
PART I 159
pang ‘shop, shopkeeper’s, store’ (as free noun ‘room’)
sik ‘style’
taykali vulgarizes nouns (as free noun ‘head [vulgar]’)
ttakci vulgarizes nouns (as free noun ‘crud’)
ttaymun ‘reason’ (quasi-free noun?)
(6) Also particle, adverb
tul ‘group’ (see detailed entry in Part II)
(7) Also postmodifier
ccay ‘and all, as it is’
chay ‘intact’
ihwu ‘after, from (the time when) ••• on’
Ilay ‘ever since, after during the past ••• ’
(8) Also postmodifier, noun
cwung < tywung ‘midst, middle of’
( -q) klm ‘impetus’ (as free noun ‘steam’)
( ~q) kyel ‘impetus’ (as free noun ‘wave’)
(9) Also postmodifier, postsubstantive
( -q) seng < ' syeng ‘quality’
(10) Also postmodifier, suffix (bound postnoun)
payngi ‘one, person’
(11) Also suffix (bound postnoun)
tali ‘one, fellow, guy’
(12) Also counter
pun ‘a portion for, enough for’ (as counter ‘minute’)
(13) Also postnominal verbal noun intransitive
chi, cha ‘bad weather (around a certain day)’
(14) Also adverb
kang ‘strong; a little over - , ••• and a fraction’
namcis, nek.nek ‘fully, all of, at least’ + number
ppa-tus ‘just under, a little short of, (falling) short’ (ppa’-tus abbr < ppanun tus)
yak ‘weak; just under, a fraction less than’ (< zyak)
(15) Also verbal noun (transitive/intransitive)
pal ‘dispatch(ed)’ [semi-literary]
(16) Also noun, adnoun, adjectival noun
•q cwung (“ccwung”) ‘a weight of (••• nyang, ton, phun); weighty’
(17) Also adnoun, numeral
pan (number +) ‘and a half’; ‘half’
(18) Pseudo-postnoun
sim (dialect variant of him) in payq sim ‘belly strength’ = ‘endurance’ or ‘greed’ and in
ip sim ‘mouth strength’ = ‘volubility’
-nyang (from ••• in yang) in i-/ku-/ce-nyang ‘this/that/that way’
Notice also the place postnouns (§5.2.3). Some of those occur as free nouns: san ‘mountain’,
kang ‘river’, ... . Others are more restricted: yang ‘ocean’, ii ‘village’, sil ‘valley’ (archaic for kol in
Omey-sil). And some are free nouns only as abbreviations: hang = hangkwu ‘port’.
nalum ‘depending on’
ppun ‘only’
tay (lo) ‘original state’
thong ‘impetus, ... ’
pakk ‘outside (of)’
( ~q) palam ‘impetus’ (as free noun ‘wind’)
Cf nyekh (p. 739)
160 PART I
A Reference Grammar of Korean
5.4.1. Postnoun/postmodifier adjectival noun (= adjectival postnoun/postmodifier).
The morpheme man plays several grammatical roles. When predicated by hata it is an adjectival
noun, but man hata is always preceded by a noun (N man hata) or by the prospective modifier (-ul
man hata, §5.4.3.2). Elsewhere I have treated man and hata as inseparable (i.e. nothing can
intervene), but for some speakers, at least, that is not quite true, for - man to hata is possible.
5.4.2. Postsubstantives.
There are four constructions that involve a substantive (-um/-m) followed by a morpheme. I treat
the morphemes that can follow as a subclass of postnoun and call them postsubstantives: -um a, -um
sey, -um say, and -umq seng. The last word (seng) also occurs as a postnoun and as a postmodifier.
For the meanings and use of the constructions, see the entries in Part II.
5.4.2.1. Postsubstantive adjectival noun.
This formidably labeled subcategory is set up to account for the peculiar behavior of the
morpheme cik in the construction -um cik ha-, for which the meaning and use will be found in Part II.
Elsewhere I treat cik and ha- as inseparable, but for some speakers the focus particle to can intervene.
5.4.3. Postmodifiers.
A postmodifier occurs after the several modifier categories of inflected words (§9.3) typically,
exclusively, or exclusively in a clearly distinct meaning. In the list below, the postmodifiers are divided
into groups according to privileges of occurrence.
(1) Exclusively after modifiers
a ‘question’
(-ulq) c e /aksimyen ‘if
ccok-ccok, cok-cok ‘every occasion that’
cii ‘uncertain fact, ... ’
«2 ‘(the time) since’
cince [obsolete] ‘behoovement’
cintay [obsolete; colloquial] ‘time when’
cuk(-sun) ‘when’
i ‘question’
ka, ko ‘question’
nawi ‘(not) enough to, ... ’
(-ulq) say [obsolete] ‘since, while’ (< s oy)
swulok ‘to the full extent that’ (< so Iwo k)
twung) ‘one of two conflicting states’
tul‘conceded fact’(< Vo/)
ya, yo ‘question’
(2) Exclusively after modifiers in the relevant shape or meaning
cwul ‘likely fact; way, ability’
II ‘definite fact; experience’
kes ‘tentative fact, ... ’
II ‘reason’ (< " LI )
pa ‘tentative fact; circumstance’
phok ‘supposition; appearance; seeming’
phum ‘appearance; behavior’
tey ‘circumstance, event’
the ‘footing, standing, relationship;
(= kyengwu) circumstance’
the, they ‘intention or expectation’
ttalum ‘only, just’
(3) Also postnoun
ccay ‘and all, as it is’
chay ‘intact, the original state’
nalum ‘depending on’
ppun ‘only’
seng ‘quality’
tay (lo) ‘original state, as is/was;
in accordance with; as soon as’
thong ‘impetus, ... ’
(4) Also postnoun, noun
klm, kyel, palam ‘impetus, ... ’
(6) Also postnoun, suffix
payngi ‘one, person’
(8) Also inflectional ending, ? particle
kwu(me)n, kwun a, kwulye ‘oh I see ... ’
(10) Also noun
kkuth ‘the final consequence’
kyem ‘at the same time’ (SEE p. 672)
(5) Also postsubstantive
(-ulq) sey - see the entry in Part II
(7) Also bound noun (after summative, p. 685)
mangceng ‘although’
(9) Also noun, adverb
han phyen ‘in addition, and, but, ... ’
nameci ‘excess, remainder’
seym ‘calculation, conjecture, speculation’
A Reference Grammar of Korean
PART I 161
(11) Also particle
(-ulq) son = s-un [obsolete] ‘[as for] the likely fact (that •••)’
A group of pseudo-postmodifiers, regular nouns (free or quasi-free), occur more widely:
cakceng ‘intention, resolve’
cek ‘time’
(-tun) cha ‘the course of’
Cham ‘the point, the verge’
checi ‘situation, circumstance’
han ‘extent, limit’
hwu ‘after’
hyencang ‘the very act / scene’
hyengphyen ‘process, circumstance’
i ‘person, one; fact, act’
kam ‘feeling’
kes ‘thing; one’
kil ‘way’
kkatalk ‘reason’
ko (lo) ‘(for) the reason’
kyengwu ‘circumstance’
Hyu ‘reason’
mal ‘words, ... ’
matang ‘instance, case’
mokcek ‘aim, purpose’
mo.yang ‘appearance’
mulyep ‘time’
nolus ‘job, role, part’; ( ~ i ‘but’)
nyekh ‘direction; toward’
pakk ‘outside of, except for’
sai, say ‘midst’
sesul ‘force’
taum ‘next’
taysin ‘substitute; instead of’
tey ‘place’ ( £ tey ‘circumstance’)
the ‘site’ ( £ the, they ‘intention,
expectation’)
thek ‘reason, grounds’
(to)cwung ‘midst’
tong-an ‘while’
ttay ‘time’
twl ‘after’
yang ‘pretense; appearance; intention’
yeyceng ‘intention’
Volyang ‘plan, intention’
5.4.3.1. Postmodifier verbal nouns intransitive inseparable (= inseparable verbal
postmodifiers).
This heavily labeled category is needed to account for chey (chek) ‘pretense’ and ssa ‘appearance’
in constructions of modifier + chey (or ssa) + ha*. These words are postmodifiers that are at the
same time also verbal nouns. For examples, see Part II.
5.4.3.2. Postmodifier adjectival nouns inseparable (= inseparable adjectival
postmodifiers).
There are four postmodifiers that are at the same time adjectival nouns inseparable:
ak/lak ‘one of two alternating states’ -q pen (or “ppen”) ‘on the verge of’
man ‘worth -mg’ (also particle, postnoun) wu ‘general appearance’
But at least some speakers allow the focus particle to between each of these and the following ha-
(Cf §5.4.1); for such speakers these words belong in the next group. (For examples, see Part II.)
5.4.3.3. Postmodifier adjectival nouns separable (= separable adjectival postmodifiers).
Each of two postmodifiers is at the same time an adjectival noun, like those in the preceding
section, but differs from them in that it can be separated from the ha-, e.g. by the focus particle to
‘even, also’ (for examples, see Part II):
pep ‘good reason to be, ... ’ tus ‘the idea/feel of’
5.4.3.4. Pre-inseparable postmodifier.
The postmodifier s%ng ‘appearance’ (< 'SYANG) is unique: in the standard language it is always
followed by the postnominal adjective inseparable siph-. For examples, see p. 773.
5.4.4. Counters. See §5.5.2.
162 PART I
A Reference Grammar of Korean
5.4.5. Bound postnouns (suffixes).
As with the adnouns (§5.3.3), some postnouns have quite stringent occurrence restrictions. Most
which are of native Korean origin have been included as “free postnouns” above. Some of those also
occur, as do a number of other morphemes, after bound elements such as verb stems, bound nouns of
various sorts, etc. For that reason there is some overlap between the earlier list and the following lists.
The suffixes in §5.4.5.1 are non-Chinese; they typically attach to native Korean elements. In the
following section are the Chinese suffixes, which typically attach to Chinese elements. For the Chinese
list, examples are provided to show why they deserve to be treated as suffixes. As with the adnouns we
are again vexed by the problem of deciding freedom and bondage of Chinese morphemes; in fact, with
the suffixes it becomes even more of a problem. Let’s consider some examples. The free noun hoy
means ‘meeting, ... ’, the free noun um.ak means ‘music’. Should we not consider um.ak hoy
‘concert’ as simply a construction of two nouns, like hak.kyo sensayng ‘school teacher’ or yeki koki-
cap.i ‘the fishing in this place’? The noun yak ‘drug, medicine’ is free and pang ‘shop’ is a free
postnoun; should we consider yak pang ‘drugstore’ as two words? Since pyeng ‘illness’ is a noun and
phipu ‘skin’ is a noun, is phipuq pyeng ‘skin disease’ a noun + noun construction? What about pyeng
cwung in the midst of illness’? In general, I have conservatively treated one-syllable Chinese elements
as essentially bound, with their freedom apparent only when they are in construction with non-Chinese
elements. So I have treated pyeng as a free noun in expressions such as kapyewun pyeng ‘a light
illness’ and palq pyeng ‘foot-soreness’, but as a bound element in cwungpyeng ‘serious illness’ where
cwung- is a bound adnoun, and in pyengwen ‘hospital’ where -wen is a bound postnoun. I am
uncertain what to do with phyeyq pyeng ‘lung disease, TB’, where phyey is a Chinese element that is
the free noun meaning ‘lungs’, yet the occurrence of the reinforcement marker -q- argues for treating
the combination as a construction of noun + noun. But -q- shows up within Chinese compounds that I
would certainly not want to treat as two words (sanqpo ‘walk’, munqca ‘written characters’, saqken
‘incident’), and there are many cases where it would be impossible for -q- to surface, e.g. before -wen.
Perhaps further studies of frequency and distribution will resolve these problems. Meanwhile, the
best we can say is as follows. Many Chinese morphemes sometimes appear, at least weakly, as free
nouns; but all Chinese morphemes are at least sometimes bound. In compounds with other Chinese
morphemes it is better to regard any one-syllable constituent as bound (bound adnoun, bound noun,
bound postnoun) unless it is clearly proven otherwise. Dictionaries of Korean sometimes hyphenate
obvious two-syllable compounds, especially when they were made up in Korea and have no
counterparts in China.
I am inclined to make a special exception of free Chinese nouns and counters + (™q) swu ‘the
number of ’ because of the exceptionally wide range of distribution (Cf §5.5.1). So I would write
chwulqsayngq swu ‘the number of births’, samangq swu ‘the number of deaths’, haksayngq swu ‘the
number of students’, ceycak swu ‘the number of products’; kwenq swu ‘the number of volumes’,
salamq swu ‘the number of people’ - but inq-swu ‘the number of people’ because in ‘person’ is not
normally used as a counter in Korean. I am also tempted to space off ••• cek ‘-4c’ and ••• sang ‘—wise’
for similar, but less compelling, reasons: see pp. 151, 440, 769. I prefer to hyphenate compounds of
one-syllable synonyms or antonyms: pok-tek ‘happiness and prosperity’, cen-hwu ‘before and after’.
5.4.5.I. Core suffixes.
The core suffixes can be divided into twelve groups:
(1) diminutives
(2) miscellaneous (pseudo-diminutives, vulgarizers, personalizers, ... )
(3) deriving both adjectival nouns and impressionistic adverbs
(4) deriving adjectival nouns
(5) deriving adverbs
(6) deriving adverbs from iterated nouns
(7) deriving adverbs from processive verbs
A Reference Grammar of Korean
PART I 163
(8) deriving adverbs from adjectives
(9) deriving nouns from adjectives or processive verbs
(10) deriving nouns from processive verbs or nouns
(11) pseudo-suffix -si deriving noun from processive verb
(12) deriving excess numeral from decimal numerals
Details on the individual items will be found in the entries of Part II.
(1) Diminutives
-k
-ak/-ek
-a(y)ki / -e(y)ki
-ang/-eng
-a(y)ngi/-e(y)ngi
-a(y)ci
-al/-el
-al.i/-el.i, (-oil/-all)
-ul%(y)ki/-l%(y)ki
-ul%(y)ngi/-l%(y)ngi
-amchi (? -* 5)
(2) Miscellaneous
-ttakci vulgarizes nouns (also postnoun, noun ‘crud’)
-ttakseni (?) vulgarizes nouns
-taykali vulgarizes nouns (also postnoun, noun ‘head’ [vulgar])
-meli vulgarizes nouns (also postnoun, noun ‘head’)
-pal vulgarizes nouns (? also noun ‘foot’)
-akw(un)i vulgarizes nouns
-ceng ‘one’ [vulgar]
-cengi ‘stuff’ [vulgar]
-tali ‘one, fellow, guy’ (also postnoun)
-takwu, -takw(un)i ‘hard thing’ (vulgarizes noun)
-(t)tayki ‘thing, one’
-te(y)ki ‘thing, one, guy’
-tek.kwungi ‘thing, one, guy’
-ttuki ‘thing, one, guy
-(t)twungi ‘thing, one, guy’
? -cwungi ‘one’
-chwungi ‘one’
-chongi ‘one’ (?)
-cha(y)ngi ‘one, thing, stuff’
-eci ‘stuff’
-thong ‘thing; part of body’
-(q-)po ‘one, thing, person’
-potwu ‘one, thing, person’
-payngi ‘one, thing, person’ (also postnoun, postmodifier)
-pangi, -pe(y)ngi ‘one, thing, person’
-eng-payngi ‘person’ (diminutive + ...)
-khengi ‘thing, person’
-kwangi ‘person’
-swungi ‘one, thing’
-eni (? -*• 5)
-wuni [dialect]
-keypi (in tes-keypi)
-(k)kayngi
-khe(y)ngi
-che(y)ngi
-ma(y)ngi
-tayngi (in yeph-tayngi)
-thayngi
-the(y)ngi
-sakwi, -say (in iph-sakwi/-say)
164 PART I
A Reference Grammar of Korean
-songi ‘person’
-soy ‘person’, makes informal names for boys
-tol(-i) ‘stone’, popular in boys’ names
-tong(-i) makes names endearing names for children
-i ‘-y, -ie’ (after consonant only) makes names endearing (children) or jocular/derisive (adults)
-huy! makes female names
-huy 2 makes pronouns plural
-a(y)mi/-e(y)mi ‘one, thing, person’; makes animal/fish/bird names
-a(y)pi /-e(y)pi makes animal/fish/bird names
-pak.i ‘an inlaid one, one with something stuck in or attached’
-pat.i ‘receptacle, ... ’
(3) Deriving both adjectival nouns and impressionistic adverbs
-(c)cok / -(c)cwuk
-mak / -mek
-(c)cak / -(c)cek, -chak / -chek
-(p)pak
-(c)cang = -(c)cak
-ppuk
-(c)cik
-(s)sek
-(c)cimak
-sil
-cin
-sin
-(c)cum
-sul
-(c)cumak
-(s)swuk
-kkak / -kkek
-swung
-(k)kis
-ttak / -ttek
-(k)kul
-ttuk
-(k)kus
-(t)twuk
(4) Deriving adjectival nouns
-kom, -(k)kum (hata) (Cf mankhum, ittakum)
(5) Deriving adverbs
-ccolok in the word amu-ccolok and its synonym md-ccolok (? < -tolok, ? < ecci ha-tolok)
-khwung
-ulu, -wulwu
(6) Deriving adverbs from iterated nouns
-i (Cf 8, 9)
(7) Deriving nouns from processive verbs
-ay (variant -ey), -kay (variant -key) ‘gadget, device, -er’
(8) Deriving adverbs from adjectives
-i/-li; -chwu, -wu
(9) Deriving nouns from adjectives or processive verbs
-i
(10) Deriving nouns from processive verbs or nouns
-wung
(11) Pseudo-suffix -si- (in nakk-si ‘fishing’ < naks- + -i = 9)
(12) Deriving excess numeral from decimal numerals
-nam.un (usually spelled -namun)
We might also add the suffix that makes approximate numerals out of numerals: -es/-e (with some
irregularities). And there are suffixes like -(u)k, -(u)l, -(u)m, -(u)n, -(u)s, ... , involved in the
derivation of impressionistic adverbs (§12).
A Reference Grammar of Korean
PART I 165
5.4.5.2. Chinese suffixes.
The Chinese suffixes (or bound postnouns) are arranged alphabetically below. Compare the lists in
Choy Hyenpay 651-2, Kim Pyengha 115-6, CM 1:218-31. Some of the morphemes could be said to be
free nouns in literary uses or in special meanings, e.g. hak = hak.mun ‘learning’. We might question
whether cung jE ‘ailment’ and seng fit ‘nature’ should not be treated as free nouns. Note that sang _h
‘-wise’ and cek ‘ ~ic’ have particularly wide combinatorial privileges in forming quasi-adnouns
(§5.3.1). Good cases could be made for including as free postnouns (§5.4) the following items: -ci
‘periodical’, -ce ‘authored by ■••’, -cok ^ ‘tribe’, -hwa ‘flower’, -kyey = kyey ^ ‘world’,
-kyo = kyo ‘religion’, -phyen %} ‘compiled by •••’, -sen ‘selected by
List of Chinese suffixes (bound postnouns)
SHAPE
MEANING
EXAMPLES
-a
5S
child
cenung — ‘feeble-minded child’, chencay — ‘child genius,
precocious child’, honhyel — ‘mixed-blood child, half-
breed’; Cf koa ‘orphan’, soa ‘infant, child’
-an
(_I_f
shore, littoral
Sehay — ‘the West (= Yellow) Sea Coast’, Tonghay — ‘the
East (= Japan) Sea Coast’
-ca
#
person, fellow
aytok — ‘devoted reader’, ko.yong — ‘employee’, 'notong
~ ‘laborer’; Cf hakca ‘scholar’, pyengca ‘invalid’
-cang]
head, chief
wiwen — ‘the head / chairman of a committee’, ... ; Cf
sacang ‘company president’
-cang 2
place
chwuk.kwu — ‘a football field’, kyengma — ‘a race track’,
sakyek — ‘a shooting ground / gallery, a firing range’,
wuntong — ‘a playground, an athletic field’; Cf nongcang
‘farm’, kongcang ‘factory’
(q)-cang 3
a*
document, letter
col.ep — ‘diploma, graduation certificate’, hyep.pak ~
‘intimidation letter’, sokayq — Tetter of introduction’,
sin.imq — ‘credentials’, wiimq — Tetter of attorney’,
kongkayq — ‘open letter’, chotayq — Tetter of invitation’
-ce
?
authored by - ,
written by -
Kim paksa — ‘written by Dr Kim’, kwahak-ca — ‘authored
by a scientist’, oykwuk-in — ‘by a foreigner’
-cek
(ft
•••iC, “0031
See Part II, p. 440; Cf p. 151.
-cel
@5
festival
along — ‘Children’s Day’, kaychen — ‘Foundation Day [of
Korea]’, sengthan — ‘Christmas’
(q)-cem
point of
chwulpalq ~ ‘starting point, point of departure’, tochak ~
‘arrival point’, wikiq — ‘point of danger’
-ceyi
remedy (for •••)
sohwa — ‘a digestant’, sotok — ‘a disinfectant’, salkyun ~
‘an antiseptic’
-cey 2
f!J
system
kyoyuk — ‘educational system’, 'yangwen — ‘bicameral
system’, 'yuk-sam-sam — ‘6-3-3 system (of schools)’,
pongken - ‘feudal system’
-chayk
m
policy
pankong — ‘anti-communist policy’, pan-Mi — ‘anti-
American policy’, yunghwa — ‘a policy of appeasement’;
Cf cengchayk ‘policy’
-che
place, office,
agency, bureau,
large facility
insa — ‘personnel office’, kongpo — ‘Office of Public
Information’, kunmu — ‘place of employment’, kwukup
— ‘relief agency’, kwan.li — ‘administrative office’,
'yen.lak — ‘liaison office’
-cheng
m
government office,
administrative center
Cwungang — ‘the Capitol’, Oyca ~ ‘Office of Foreign
Supply’; Cf Slcheng ‘City Hall’
166 PART I
A Reference Grammar of Korean
-cho
W
grass, weed, plant
kumcam ~ ‘dandelion’ (= mintulley), manpyeng ~ (a
rhododendron), kumpul ~ ‘elecampane (Inula japonica)’,
pullo ~ ‘a herb of eternal youth’; CFWncho ‘orchid’
-cii
m
place, land
kecwu ~ ‘place of residence’, cem.lyeng ~ ‘occupied
territory’, cwuthayk ~ ‘a residential area’; Cf koci
‘upland’, phyengd ‘flatland’
-ci 2
1 paper
2 newspaper titles
insway ~ ‘printing paper’, panca ~ ‘ceiling paper’,
hwasen ~ (? hwa-senci) ‘a thin rough paper’, pyek ~ =
topay ~ ‘wallpaper’ [toypay (CM 1:225) must be dialect]
Ppulawuta/Ppulaputa ~ ‘Pravda’, Icupeycciya ~ ‘Izvestia’,
Thaimucu ~ ‘the Times’
-ci 3
DJU'
periodical
cwukan ~ ‘a weekly’, welkan ~ ‘a monthly’; Thaimu ~
‘Time Magazine’, Laiphu ~ ‘Life Magazine’
-cok
m
tribe, group, people,
nationality
Intian ~ ‘Indian tribes’, Thipeythu ~ ‘Tibetans’, n Yecin
~ ‘the Ju(r)chen’
-cong
m
variety
kaylyang — ‘improved variety’, caylay ~ ‘native variety’,
oylay ~ ‘nonnative variety’
(q)-cung!
certificate
hapkyek — ‘certificate of qualification (passing)’, itongq ~
‘certificate of moving’
(q)-cung 2
m
ailment
hyenhwunq ~ = ecilq ~ ‘vertigo, dizziness’, kyelhayk ~
‘tuberculosis’, pokchangq ~ ‘swollen-belly ailment’,
sinkyengq ~ ‘nerve disorder’; emseyq ~ ? ‘depression;
pessimism’; Cf kalqcung ‘thirst’, hwa( 7 q)cung ‘anger,
displeasure’. This element also occurs bound in silhq-cung
/silccung/ ‘displeasure’ (< adj silh-) and kapkap-cung
‘uneasiness’ < adj-n kapkap (ha-).
-cwu
master, boss
kongcang ~ ‘factory boss’, kiep ~ ‘boss of the enterprise’
-e
=E
on
language,
Cwungkwuk — ‘Chinese’, Hankwuk ~ ‘Korean’, Upon ~
word(s)
‘Japanese’, hoching ~ ‘designation(s)’, swusik ~ ‘a
modifier’; Cf en.e ‘language’, Yenge ‘English’, kwuk.e
‘vernacular’; cwue ‘subject’, swul.e ‘predicate’
-ha
T
under
SEE Part II, p. 514.
-hak
m
science, study,
sahoy ~ ‘sociology’, kyengcey ~ ‘economics’, mulli ~
-ology, -ics
‘physics’, en.e ~ ‘linguistics’
-han
m
person, guy,
muloy ~ ‘shifty loafer’, mun.oy ~ ‘layman’, putek ~
fellow
‘unvirtuous fellow’; Cf koyhan ‘suspicious-looking guy’
-ho
1 number
cey sip-sa ~ (sil) ‘(Room) No. 14’
2 issue, number
kinyem — ‘commemorative issue/number’
3 name, designation
Masan ~ ‘the S.S. Masan’, Thongil ~ ‘the Unification
Express’ (a train); Cf kwuho ‘slogan’
-hoy
EJ
gathering,
cwatam ~ ‘a roundtable discussion’, ‘nangtok ~ ‘a
it
meeting
(gathering for) reading; a reading group’, um.ak ~ ‘a
concert’; CFmyenhoy ‘interview’
-hwai
conversion,
hap.li ~ ‘rationalization, streamlining, reordering’, kikyey
•••ization, -ize
~ ‘mechanization’, kwuk.yu ~ ‘nationalization’, mincwu
~ ‘democratization’, tongmul ~ ‘brutalization’, tosi ~
‘urbanization’
-hwa 2
It
flower
haytang ~ ‘sweet briar’, mukwung ~ ‘the Rose of
Sharon’, chaysong ~ ‘portulaca (rosemoss)’, nungso -
‘trumpet flower’; Cf kwuk.hwa ‘national flower’
A Reference Grammar of Korean
PART I 167
-il
B
day
(also counter)
-in
A
person
-kai
M
professional
-ka2
m
song
-ka3
-ka 4
pj
Dj
m
quasi-title
street
-kam
!§
a feeling
-kayk
&
guest, person
-ki.
§5
device,
instrument
-ki 2
m
1 machine
-ki 3
m
2 (air)plane
period of time
-kong
X
artisan
(q)-kwa
n
1 course, class
2 taxonomic family
3 office, bureau,
section
-kwari)
-kwan 2
m
government
official
place, building
(q)-kwen f|| power, authority,
(also noun)
konghyu ~ ‘a legal holiday’, kwukchi ~ ‘National
Humiliation Day’, thansayng ~ ‘birthday (of a sage)’; Cf
sayngil ‘birthday’, kiil ‘fixed date; death anniversary’
ca.yen ~ ‘a natural person, natural man’, Hankwuk ~ ‘a
Korean’, munhwa ~ ‘a person of culture’; Cf siin ‘poet’
cengchi ~ ‘politician’, mlswul ~ ‘artist’, sap.hwa ~
‘illustrator’, thamhem ~ ‘explorer’, yeyswul ~ ‘artist’;
Cf cak.ka ‘writer’
aykwuk ~ ‘patriotic song’, cacang ~ ‘lullaby’, nongpu -
‘farmer song’; Cf coka ‘dirge’, songka/sengka ‘hymn’
affixed to surname (humble or pejorative)
cwuthayk ~ ‘residential street’, Cong.lo sam ~ ‘Bell Street
at Third Street’ (an area once notorious for prostitution),
penhwa ~ ‘busy street, thoroughfare’
ap.pak ~ ‘oppressive feeling’, kincang ~ ‘tense feeling’,
mancok ~ ‘a feeling of satisfaction’, pul .an ~ ‘uneasy
feeling’; Cf chok.kam ‘the sense of touch’, yuk.kam
‘sensuality’, ’yuk.kam ‘sixth sense’
mangmyeng ~ ‘an exile, a refugee’, mohem ~ ‘an
adventurer’, wuntong ~ ‘a sport spectator, fan’; Cf
hayngkayk ‘tourist’, sungkayk ‘passenger’
kyeylyang ~ ‘ga(u)ge, meter’, chuk.lyang ~ ‘surveying
instrument’, chuk.wu ~ ‘a rainfall ga(u)ge’ pyen.ap ~
‘transformer’, punto ~ ‘protractor’; sohwa ~ (1) ‘fire-
extinguisher’, (2) ‘digestive organs’; Cf hyungki ‘lethal
weapons, arms’
apchak ~ ‘press’, cwuco ~ ‘type-caster’, insway ~
‘printing machine’, palqtong ~ ‘motor’
phok.kyek ~ ‘bomber’, swusong ~ ‘transport plane’
chochang ~ ‘pioneer days’, pun.lan ~ ‘chaotic period’;
Cf choki ‘early period’, malki ‘later period’
klnung ~ ‘technician’, kumsok ~ ‘metal worker’, mophi-
ceyphum ceyco- ~ ‘furrier’, pangcik ~ ‘textile worker’,
pelmok ~ ‘lumberjack’, swuk.lyen ~ ‘skilled craftsman’;
Cf mok.kong ‘woodworker’, "yekong ‘factory girl’
kwuk.eq ~ ‘Korean course’, swuhak ~ ‘mathematics
course’, Yengeq ~ ‘English course’; Cf hak.kwa ‘course’
cangmiq ~ ‘roses’, clntallayq ~ ‘azalea’
cengpoq ~ ‘intelligence/information bureau (or section)’,
hayngcengq ~ ‘administrative office’, insaq ~ ‘personnel
division/office’, pokup ~ ‘supply section’
canghak ~ ‘an inspector of schools’, kemchal ~ ‘public
prosecutor’, kem.yel ~ ‘censor’, sihem — ‘examiner’
mlswul ~ ‘art gallery’, pak.mul ~ ‘museum’, slkong ~
‘public auditorium’, simin ~ ‘City Center’, taysa -
‘embassy’, yenghwa ~ ‘movie theater, cinema’;
Cf hoykwan ‘meeting hall’, ^ekwan ‘hotel’
hayngcengq ~ ‘administrative authority’, myeng.lyengq ~
‘commanding authority’, sapep ~ ‘judicial power’; Cf
silqkwen ‘real power’, phayqkwen ‘hegemony’
168 PART I
A Reference Grammar of Korean
-kwui
P
entrance, wicket,
hole, opening,
window
cepswu ~ ‘reception window’, chwul.ip ~ ‘entrance (and
exit)’, chwul.nap ~ ‘window/wicket for collections and
disbursements’, kayphyo ~ ‘the ticket (fare) adjustment
window’, punhwa ~ ‘a (volcanic) crater’; Cf ipkwu
‘entrance’, chwulkwu ‘exit’
-kwu2
B
7X
tool, implement
munpang ~ ‘stationery supplies’, panghan ~ ‘cold-weather
gear’, wuntong ~ ‘athletic goods’; Cf kikwu ‘utensil,
appliance’, tokwu ‘tool’
J
i
m
country, state,
nation
konghwa ~ ‘republic’, mincwu ~ ‘democracy’, kwuncwu
~ ‘monarchy’; Cf ponkwuk ‘homeland’, cek.kwuk
‘enemy country’, akwuk ‘our country’
-KWUK2
agency, office
chelqto ~ ‘railway station’, sam.lim ~ ‘bureau of forestry’,
wuchey ~ ‘post office’; Cf yak-kwuk ‘pharmacy’
-kwun
army
cengpu ~ ‘the government forces’, hyek.myeng ~ ‘revolu¬
tionary army’, haypang ~ ‘an army of liberation’; Cf
haykwun ‘navy’, kongkwun ‘air force’
-kyeyx
n
world, circles,
kingdom,
-dom, realm
chwulphan ~ ‘publishing circles’, sasang ~ ‘the world of
ideas’, tongmul ~ ‘the animal kingdom’, um.ak -
‘musical circles’; pi-Kathollik ~ ‘non-Catholic circles’
-kyey 2
&
It
system; lineage;
faction
tongmul ~ ‘animalia, the animal kingdom’, thayyang ~
‘solar system’
-kyey 3
ga(u)ge, meter;
scheme
ap.lyek ~ ‘manometer, pressure gauge’, cheyon ~ ‘(body)
thermometer’, han.Ian ~ ‘(weather) thermometer’, kiap ~
= chengwu ~ ‘barometer’, phungsok ~ = phung.lyek
~ ‘anemometer’; miin ~ ‘ensnaring with a beautiful
woman’; CFsikyey ‘timepiece’
-kyey 4
report
chwulqsek ~ ‘attendance report’, kyelqsek - ‘report of
absences’, chwulqsayng ~ ‘birth report’, kecwu ~ ‘report
of residence’
-kyo
religion,
teaching
Isullam ~ ‘Islam’, Kitok ~ ‘Christianity’, Molumon ~
‘Mormonism’; Cf Pulkyo ‘Buddhism’
-lo
n
street
Seycong - ‘Seycong Street’, Thayphyeng ~ ‘Thayphyeng
Street’, Ulqci ~ ‘Ulqci Street’; Cf tolo ‘roadway, street’
-lyo
m
charge, fee;
materials
See Part II, p. 679.
-lyu
m
kind, sort,
species
inkan ~ ‘human species’, nuktay ~ ‘wolf species’, pho.yu
~ ‘mammalia’, tongmul ~ ‘animal species’; Cf alyu
‘adherent, follower; a second’, cong.lyu ‘kind, sort’
-mang
m
network
chelqco ~ ‘barbed wire’, cocik ~ ‘organization(al)
network’, kyoyang ~ ‘cultural network’, 1 y en -l a ^ ~
‘communications network’
-mul
stuff, thing,
matter
chwulphan ~ ‘publications’, insway ~ ‘printed matter’,
paysel — ‘excrement(s)’
-pay
m
people, group
[pejorative]
cengsang ~ ‘petty politicians; politicos’, kangto ~
‘robbers’, kansang ~ ‘fraudulent merchants’, moli -
‘profiteers’, sonyen ~ ‘young people’
-pha
M.
group, faction,
clique
cenhwu ~ ‘the apres-guerre (postwar) group’, insang ~
‘the impressionists’, ] nangman ~ ‘the romantics’; Cf
sinpha/kwupha ‘the new /old school’
A Reference Grammar of Korean
part i 169
-phum goods
-phung
style,
manner(s)
-phyen
m
compiled by -
(abbr <
phyen[chan] ham)
-pi
m
expenditures for -
-Pi*l
sb
section;
office;
ministry
-pu 2
menial; workman
-pyel
ffl
division,
separation,
classification
(q)-pyeng
illness
-S3]
m
person,
master
-sa 2
±
scholar,
person
-sa3
it
company,
corporation
-sa4
£
history of -
-sayng
£
1 student
2 birth
-se
*
writing,
document
cayko ~ ‘(goods in) stock’, hapkyek ~ ‘approved goods’,
kakong ~ ‘processed goods’, swuip ~ ‘imported goods’,
wulyang ~ ‘superior merchandise’; Cf kyengphum ‘a
premium (free gift)’, sangphum ‘merchandise’
cangkwun ~ ‘proud manner’, cangpu ~ ‘manly manner’,
se.yang ~ ‘western (= Occidental) manners’, sikol ~
‘country manners’, Kwantong ~ ‘Kwantong style’
Hankul Hak.hoy ~ ‘compiled by the Korean Language
Society’, Mlswul Hak.hoy ~ ‘compiled by the Art
Institute’, Mun Seyyeng ~ ‘compiled by Mun Seyyeng’;
Cf phyenca ‘compiler’
chilyo ~ ‘medical expenses’, kasel ~ ‘construction costs’,
saynghwal ~ ‘living expenses’, swusen ~ ‘(expenditures
for) repairs’; Cf hoypi ‘membership fee/dues’
cayceng ~ ‘ministry of finance’, cheyyuk ~ ‘department of
physical education’, wisayng ~ ‘ministry of public health’;
Cf ponpu ‘headquarters’
chengso ~ = socey ~ ‘cleaning man, janitor’, chwlsa ~
‘cook’, seythak ~ ‘laundryman’; Cf hwapu ‘fireman,
stoker’, kwangpu ‘miner’
chwulqsin ~ ‘classification by place of birth’, cik.ep ~
‘breakdown by occupation’, kyeykup ~ ‘class division’,
namnye ~ ‘separation by gender’, sengcek ~ ‘grouping
by grades (achievement)’, sengpun ~ ‘classification by
elements (components or ingredients)’, n yen.lyeng ~
‘division by age’; Cf phanpyel ‘discrimination’
phipuq ~ ‘a skin disease’, simcangq ~ ‘heart trouble’,
wicangq ~ ‘alimentary disorder’; Cf phungqpyeng ‘palsy,
paralysis’
^palq ~ ‘barber’, maswulq ~ ‘magician’, senkyo ~
‘missionary’; Cf klsa ‘technician’, kyosa ‘teacher’, uysa
‘physician’
kikwan ~ ‘engineer’, pihayng ~ ‘aviator’, pyenho ~
‘lawyer’; Cf paksa ‘Ph.D.’, haksa ‘B.A., A.B.’
chwulphan ~ ‘a publishing house, a publisher’, sinmun ~
‘a newspaper (company)’, thongsin ~ ‘a news agency’;
Cf hoysa ‘company’
kenchwuk ~ ‘the history of architecture’, Mikwuk ~
‘American history’, munhak ~ ‘history of literature’,
se.yang ~ ‘history of the west’; Cf *yeksa ‘history’
chonyen ~ ‘freshman’, kangsup ~ ‘short-course student’,
silqsup ~ ‘trainee’, yenkwu ~ ‘research student, student
researcher’; Cf haksayng ‘student’
yun.welq ~ ‘born in a leap-month’, Im-cin — ‘born in the
29th year of the 60-year cycle’; Cf chwulqsayng ‘birth’
cungmyeng ~ ‘certificate’, *ilyek ~ ‘a personal history,
one’s (career) resume, vita’, incung ~ ‘(a written)
authentication’, pocung ~ ‘(written) guarantee’; Cf congse
‘vertical writing’, hoyngse ‘horizontal writing’
170 PART I
A Reference Grammar of Korean
-sel
u
theory, view
pantay ~ ‘opposite view’, 'yenghon-pulmyel ~ ‘the theory
of eternal life’; Cf haksel ‘scholarly theory’
-sen!
II
line
cehang ~ ‘line of resistance’, Kyeng-Pu ~ ‘Seoul-Pusan
line’, pangwi ~ ‘defense perimeter’, samphalq ~ ‘the
38th parallel’; Mayk-Ate ~ ‘the MacArthur line (along the
'Naktong River)’; Cf congsen ‘vertical line’, hoyngsen
‘horizontal line’
-sen 2
m
selection;
selected by -
Kim Caywen ~ ‘selected by Kim Caywen’, slmsa wiwen ~
‘selected by a judging committee’
(q)-seng
14
nature,
quality
chenyeq ~ ‘virginity’, chwungsilq ~ ‘substantiality;
loyalty, faithfulness’, pyenthayq ~ ‘abnormality’, thuk.iq
~ ‘peculiarity’, wihemq ~ ‘dangerousness’; Cf
phumseng ‘quality of goods’. See Part II, p. 773.
-sil
room, office,
lab, small
institution
cokak ~ ‘sculptor’s studio’, mok.yok ~ ‘bathroom’,
silhem ~ ‘laboratory’, tose ~ ‘the library (room)’, ungcep
~ ‘the drawing-room’, yenkwu ~ ‘seminar (room)’; Cf
kayksil ‘guestroom’, onsil ‘hothouse’
-so
P/T
place,
institution,
institute,
facility
chilyo ~ ‘infirmary’, HpaHq) ~ ‘barber shop’, samu ~
‘office’, yenkwu ~ ‘research institute (facility, laboratory)’,
Voyang ~ ‘sanatorium (sanitarium)’; Cf cangso ‘place’,
cwuso ‘residence’
-swu
hand, person
kikwan ~ ‘locomotive engineer’, kyohwan ~ ‘switchboard
operator’, wuncen ~ ‘driver’; Cf coswu ‘assistant’, kiswu
‘assistant engineer’, senswu ‘athlete’
-swul
#T
technique,
art, trick
insway ~ ‘the art of printing’, sakyo ~ ‘the art of social
intercourse’; Cf maswul ‘magic’, swuswul ‘operation’
-tam
ok.
talk(s), tale,
report (on - )
kyenghem ~ ‘a story of personal experience’, mohem ~
‘an adventure story’, palkyen ~ ‘a tale of exploration’,
'yehayng ~ ‘a travelog’; Cf hoyhwa‘conversation’
-tayi
w
group, outfit
kyengkwan ~ ‘police squad, posse’, kyengpi ~ ‘garrison’,
thamhem ~ ‘expedition, exploration party’, ungwen ~
‘cheerers, rooters; reinforcements’; Cf putay ‘detachment’,
kwuntay ‘troops’
-tay 2
m
belt; zone
hwasan ~ ‘volcanic zone’, kwuco ~ ‘buoy, life preserver’,
sam.lim ~ ‘forest zone’; Cf citay ‘zone, belt’
-thong
m
pain, ache,
-algia
hyungpu ~ ‘chest pain’, sinkyeng ~ ‘neuralgia, nerve
pain’; Cf chithong ‘toothache’, twuthong ‘headache’
-tO!
fS.
(year) period
cak.nyen ~ ‘last year’, kumnyen ~ ‘this year’, 1960-nyen
~ ‘the year I960’; Cf "yento ‘year period’
We could treat -nyen-to as a binom -nyento.
-to 2
n
painting,
drawing, view
cokam ~ ‘bird’s-eye view’, miin ~ ‘portrait of a beauty’,
sanswu ~ ‘landscape’, tanmyen ~ ‘cross-sectional view’;
Cf cito ‘map’, chwukto ‘reduced drawing’
-weni
m
institution
koa ~ ‘orphanage’, Haksa ~ ‘the Scholars Institute (of
Kolye times)’, swuto ~ ‘monastery’, tayhak ~ ‘graduate
school’; Cf hak.wen ‘the academy’, pyengwen ‘a hospital’,
sawen ‘a temple’
-wen 2
garden; park;
institute
kwaswu ~ ‘an orchard’, tongmul ~ ‘a zoo’, yuchi ~
‘kindergarten’; Cf kongwen ‘a (public) park’
A Reference Grammar of Korean
PART I 171
-wen 3
t —1
K
clerk, member,
employee
centhwu — ‘combatant’, congep — ‘employee’, swukcik ~
‘night-duty man’, tayuy — ‘congressman’; Cf cem.wen
‘store clerk’, hoywen ‘member’, Im.wen ‘staff member’,
puwen ‘a member of the section’, sengwen ‘a constituent
member’
-ye
m
with excess,
••• odd, over
Added to decimal and higher-unit Chinese numerals;
see §5.5.2.
(q)-yem
&
inflammation,
—itis
kwancelq — ‘arthritis’, mayngcangq — ‘appendicitis’,
'nuk.makq — ‘peritonitis’
On the irregular ph y ey‘yem ‘pneumonia’, see p. 16.
-yong
m
for the use of -
haksayng — ‘for students’, kaceng — ‘for household use’,
kyosa — ‘for faculty’, namca — ‘for men’, philki — ‘for
writing (purposes)’, samu — ‘for business (use)’
See also proper names, §5.2.3. Free nouns sometimes occur in compounds as if suffixes. It is not
always easy to decide whether a single Chinese morpheme occurs alone as a free noun (other than as
an abbreviation of a binom) or not. And some bound postnouns occur only in vocabulary that is highly
circumscribed, e.g. -tam (3 ‘lake’ in Payk.lok-tam £3 m w. c White Deer Lake”), the name of the
crater lake on top of Mt Hanla on Ceycwu island.
5.5. Numbers.
Numbers are a way of quantifying things. Languages express quantification in various ways, often
as adnominal modification of the noun (“two candies”) or of a representative counter (“two pieces of
candy”), and sometimes as a noun substitute (“I want two/lots [of them]”). Some languages are more
rigid than others in the structures they permit. Korean is fairly flexible, but some of the possible
constructions are more common than others, and when two or more structures are allowed each may
have specific connotations.
5.5.1. Number constructions.
As in many languages, the number expressions in Korean introduce special problems. We
recognize two important classes of words: NUMERALS, a subclass of noun, and counters, a subclass
of postnoun. A counter occurs typically after a numeral, but it can also form a construction with the
postcounter (-q) swu ‘the number of - ’: (chayk) kwenq swu ‘the number of books’. (Chayk swO is
also said, but less commonly.) Notice that counters, as counters, are not modified by i ‘this’, ku ‘that’,
kulen ‘such’, ... . In i pen kaul ‘this autumn’, pen is a noun ‘time’.
There are three kinds of counters: UNIT, measure, and numeral. A unit counter counts
individual instances of a countable noun: chayk han kwen ‘one book’, kay twu mali ‘two dogs’, pay
sey chek ‘three boats’, sengnyang-kaypi (= sengnyang) sey kay ‘three matches’. A measure counter
registers the amount of a measurable noun (chan han can ‘one cup of tea’, maykcwu twu pyeng ‘two
bottles of beer’, sengnyang sey kap ‘three boxes of matches’) or of units of time (han si ‘one o’clock’,
han sikan ‘one hour’, han tal or il-kaywel ‘one month’, sam-nyen ‘three years’) or of money (chen
wen ‘a thousand wen’). A numeral counter is a numeral that is itself being counted, and so functions
as a counter after another numeral: sam-payk ‘three hundred’, sa-chen ‘four thousand’. Many of the
measure counters (‘cupful, boxful’, ... ) and a few unit counters (salam ‘person’ in haksayng han
salam ‘one student’) could be labeled “temporary counters” since they occur also as free nouns, often
counted by other counters: can han kay ‘one cup’ - but salam han salam (or han myeng) ‘one
person’. Some of the other counters occur also in constructions other than numbers, for example chay
(counter for buildings) in salang chay ‘detached house’, an chay ‘main house’.
Among the countable nouns, there are some that have specific unit counters but many others lack
specific counters and are counted simply by the numeral alone. The numeral without a counter can be
172 PART I
A Reference Grammar of Korean
used to count any noun. The following sentences, meaning ‘One book exists’ = ‘We’ve got one/a
book’, illustrate the constructions that occur with countable nouns that involve the nominative case
particle (Pci), numeral (Num), numeral + counter (Num-Count), and the adnominal particle (uy):
(1) N Pci Num-Count
(2) N Num-Count Pci
(3) N Pci Num-Count Pci
(4) Num-Count uy N Pci
? (4a) Num-Count N Pci
(5) N Num Pci
(6) N Pci Num
(7) N Pci Num Pci
? (8) Num uy N Pci
(8a) Num N Pci
Chayk i han kwen iss.ta.
Chayk han kwen i iss.ta.
Chayk i han kwen i iss.ta.
Han kwen uy chayk i iss.ta
? Han kwen chayk i iss.ta.
Chayk hana ka iss.ta.
Chayk i hana iss.ta.
Chayk i hana ka iss.ta.
? Hana uy chayk i iss.ta.
Han chayk i iss.ta.
Some nouns, however, do not occur in constructions of type (8a). You can say thokki hana ‘a
rabbit’ and talk (i) hana ‘a chicken’ but not *han thokki or *han talk. Instead you say han mali
thokki and han mali talk; Cf CM 1:139. With a juncture between, it is possible to say han 1 thokki
in the meaning ‘a certain rabbit’. The juncture may be hard for the ear to catch in that phrase but it
should be clear in similar phrases: han I talk ‘a certain chicken’ will have a slightly aspirated (and
certainly voiceless) articulation of Itl, while the unacceptable *han talk would have the voiced
allophone [d]. There is a distinction between han I salam ‘a certain person’, where the length of the
first-syllable vowel is maintained after the juncture, and han salam ‘one person’, where the lack of
juncture leads to suppression of the vowel length, though there may be speakers who retain the length
(Cf han kwen, ... ). Structure (8a) was quite common in the earlier language: "ney a'to'li (1459 Wei
2:6b) ‘the four sons’; "sey ' sa 'li (1445 'Yong 89) ‘three arrows’; hon the 'li lol (1447 Sek 6:27a) ‘one
hair’; "ney polo'm ay (1447 Sek 24:20b) ‘on the four walls’.
Tsukamoto 1986 seems to disallow (4a), the reduction of (4), but he is thinking of a different
source for the surface structure: a preposing of the adverbialized number rather than a reduction of the
adnominalized number. There are a few examples of the latter from earlier Korean: "twu "nas "twon
i'Gwo ( ? 1517- Pak 1:52a) ‘it is (= costs) two coins, and ... ’; nay syel'hun lyang un i i'sye ’yla
( ? 1517- Pak 1:62a) ‘I have thirty taels of silver (= money)’. And an example with an adnoun
modifying the noun: na ’y "twu swang sayhwe'lul ta ka "ta tonnye hoy 'ya po'likwa ’ la ( ? 1517-
Pak 1:35a) ‘I took my two pairs of new shoes and wore them both out getting about!’ Corresponding to
the unreduced (4) is the MK structure Num Count s N, as in hon ' cwuls kul (1481 Twusi 21:25b) ‘a
single line (of news)’ and “ ney ka'ci s "ssrwuw -"khwo lol (1447 Sek 6:4a) ‘the Four Miseries’.
The most common structure in modern Korean is that of (2) and (5): N Num(-Count) Pci. This
seems to be an inversion of (4a), the questionable reduction of (4). Since the number word modifies the
noun, we expect it to precede the noun, so that (4) is the logical starting point for deriving the other
structures. In Japanese when the noun is subject or object the most common and least “marked”
structure is (1), which adverbializes the quantifier. This structure also occurs in Korean, but it may be
a modern innovation, perhaps taken from Japanese usage, since there are no Middle Korean examples.
Japanese permits the adverbial to be preposed (put before the noun), in what Tsukamoto calls
“quantifier forward floating”, but Korean does not permit sentences like *Sey myeng chinkwu ka wuli
cip ey wass.ta *- Chinkwu ka sey myeng wuli cip ey wass.ta ‘Three friends came to my house’.
Modern examples of structures (1) and (6), with the adverbialized number:
Sacen ul han kwen mantulq yang ulo caylyo lul mouko iss.ta ‘I am gathering data with a view
to compiling a dictionary’.
Namuq kaci lul hana kkekk.ess.ta ko yatan hana namuq kaci ka elma ’na khulq sey mal ici
‘He is making such a fuss over the branch I broke, but I ask you, how big a branch is it anyway?!’
A Reference Grammar of Korean
PART I 173
’Naynyen imyen catong-cha lul hana sakey ccum toylq key ’ta ‘Next year I’ll be in a position
to buy a car’.
Helum han cip ina-ma nay cip ul hana kacyess.umyen coh.keyss.ta ‘I wish I had a house of
my own, however humble it might be’.
The unusual structures of (3) and (7) above mark both the noun and the numeral(-counter) phrase
with the nominative particle. The accusative particle, too, permits such structures (chayk ul hana lul
pwass.ta) but they are not usually compatible with other particles, such as that marking the indirect
object: *chinkwu eykey twul eykey cwuess.ta -*■ chinkwu twul eykey (or twu chinkwu eykey)
cwuess.ta ‘gave it to two friends’. But when the dative phrase is optionally marked by the accusative
particle (ul / lul) instead of the dative particle (eykey or hanthey) the structures are acceptable, at least
to some speakers: Emeni ka ai tul ul motwu lul senmul ul cwuess.ta ‘The mother gave a present to
each of the children’ (LR 24:174:n6). There are advantages to “copying” the nominative or accusative
marker instead of letting the quantifier stand as an adverb, in that the reference of the adverbialized
number could be either to the subject or to the object, so that (as Gerdts 1985 points out) A ka B
eykey C lul seys cwuess.ta is ambiguous as to whether three of the A or three of the C are involved,
whereas A ka B eykey C lul seys ul cwuess.ta is unambiguously ‘A gave three of C to B’ and A ka B
eykey C lul seys i cwuess.ta is unambiguously ‘Three of A gave C to B’.
Numbers (whether numeral + counter or just numeral) are allowed to “float” away from the
nouns they are counting when those nouns are subjects or direct objects. The float is normally not
permitted if the noun has some other role in the sentence, unless that role is secondary to an underlying
role as subject or object, as in the causative structure of Nay ka haksayng eykey seys tte-nakey
hayss.ta ‘I let three of the students leave’ *- Haksayng i seys (i) tte-nass.ta ‘Three of the students
left’. A special case is found in Nay ka haksayng eykey seys i tte-nakey hayss.ta ‘I let the students
leave in groups of three (or as a group of three)’, where the nominative-marked quantifier is allowed to
float although the underlying subject to which it refers has been converted from nominative to dative
(haksayng i -*■ haksayng eykey) under the causativization. We know that this is the nominative-
marked quantifier (and not, say, a variant of se-i ‘three persons’) because the suppletive alternant of
the marker appears in Nay ka haksayng eykey yel hana ka tte-nakey hayss.ta ‘I let the students leave
in groups of eleven (or as a group of eleven)’. The underlying structure seems to be something like: “I
let the students do it such that three [of them] leave”. In a simpler sentence without the causativization
you might get haksayng i seys i ... or haksayng ul seys ul ... ‘three of the students [as subject or
object]’. Perhaps these are just cases of a kind of pseudo-float using the multiple-case marking that is
permitted for genitives, from an underlying structure *haksayng uy seys ‘three of the students’, but
that explanation seems disconfirmed by the fact that, unlike Japanese, Korean does not permit the
structure *N uy Num. (Cf Tsukamoto 1986. I find no examples of *N s Num in earlier Korean.)
The floating of the numbers is usually called “quantifier float”, since in other languages (such as
Japanese) there are quantifiers like ‘all’ and ‘lots’ that can behave the same way as the number words.
It should be noted that Korean ta ‘all’ and manhi ‘much/many’ are adverbs, unlike the number words,
so that they will occur only in structure (1) as N pci ta/manh.i ‘N entirely/muchly’, and only that
structure is therefore found for them in earlier Korean, too. Exceptionally the adverb motwu ‘all; each,
every’ is also now (and perhaps newly) treated as a noun that can take the nominative and accusative
markers. To say ‘all N’ or ‘many/much N’ you use the corresponding adnominal forms motun and
manh.un. The word meych < myech ‘how many; a few’, is a numeral, and like the other numerals it
can stand as subject or object. The most frequent occurrence is before a noun or a counter: meych
tal(q tong-an) ‘how many months (time)’, meych salam/pun ina ‘about how many people’, meych
pen ‘how many times’. But it can also occur alone without a counter: Swu ka meych ina toysinun ci
yo ‘May I know your age?’ And, with or without the counter, it can occur in the various structures
open to the numerals: Thokki meych (mali) eykey punpay han seym in ya ‘How many rabbits do you
figure got their rations?’; Son nim meych pun i osip.nikka ‘How many guests are expected?’; Chayk
174 PART I
A Reference Grammar of Korean
meych kwen ul ilk.ess.ta ‘I read a few books’; Ku ttolay lul meych kay te sa ’ta cwuo ‘Buy a few
more of that size’; Ko ttolay meych i chac.e wass.ess.ta ‘A group (of boys) of that age had been here
to call’.
Once the quantifier is floated as an adverbial it has the freedom of other movable adverbs and may
move away from the noun to which it refers. According to Gerdts (1985:55) Ku cik.kong i sonq-kalak
i kikyey ey seys i callyess.ta can be taken either as ‘Three fingers of the workman were cut on [cut off
by] the machine’ or ‘The fingers of three workers were cut on [cut off by] the machine’.
Although the plural particle tul can freely occur after just about any phrase, and can be inserted
repeatedly to increase the emphasis, it is not quite the same as quantifier float, because the reference is
only to the subject, and that may be implied rather than expressed. Notice that (Wuli ka) chayk tul
(ul) tul ilk.ess.ta ‘We read our books’ only the second tul can be the plural-subject marker, the first
must be the postnoun marking a noun as explicitly plural. The sentence Chayk tul ilk.ess.ta is
ambiguously ‘I read the books’ or ‘We read the book(s)’ unless the accusative marker is explicitly
located: Chayk tul ul ilk.ess.ta ‘I (or we) read the books’, Chayk ul tul ilk.ess.ta ‘We read the
book(s)’.
5.5.2. Numerals.
We could define a numeral as any noun that answers the question Meych in ya ‘How many is
it?’, but we want to include a few additional items. Not only does the numeral freely occur before the
copula ita (‘it is such-and-such a number’) and before particles, both in arithmetic statements and as a
substitute for constructions of numeral + counter, but it also occurs as an adnoun before a noun or a
counter. And it appears in absolute constructions, as an adverbial phrase.
There are subclasses of numerals:
(1) quasi-numerals elma ‘how much; some amount’
meych < myech ‘how many; some’
'yang ‘both’ (adnoun only)
swu ‘a number of; some, several’*
*As an adnoun. In this use swu - is largely limited to Chinese counters, for the other
counters prefer yele - ‘several’ (< yeles ‘about ten’), but swu ••• is an option for certain
common counters: kay swu/yele mali ‘several dogs’. The morpheme is also used as a
postcounter ‘the number of’.
(2) numerals proper, core hana/han < honahlhon ‘one’,
twul/twu < "twulhl "twu ‘two’,
seys/sey/sek/se < "seyhl "sekl "se ‘three’, ...
(3) numerals proper, Chinese il ‘one’, I ‘two’, sam ‘three’, ... ;
*yeng, kong ‘zero’; pan, celpan ‘half’
(4) approximate numerals, core twues/twue < "twu'zehl "twu ze ‘about two’
(based on the bound counter -es) (‘two or more’ 1887 Scott 97), ... ;
yeles/yele < ye'lehtye'le ‘ten or so; a
number (of), quite a few’
(5) approximate numerals, Chinese il(q)-I ‘one or two’, I-sam ‘two or three’, ...
(6) excess numerals, core ye-nam(.)un/-nam(.)u ‘10-odd’ [dialect variant
(the tens + suffix -nam.un ‘left yelamu(n)], ... , ahu-nam(.)un/-nam(.)u ‘90-odd’
over’, often spelled -namun)
(7) excess numerals, Chinese sip-ye ‘10-odd’, ... , payk-ye ‘100-odd’, ... ;
(the tens and higher units + Cf mulye payk ‘no less than 100’, mulus payk
suffix -ye) ‘a hundred or so’
Note that - (k)awus and ••• pan ‘and a half’ are postnouns that appear after the construction
numeral + counter. CFthe numeral pan ‘half’.
The two sets of numerals proper, core and Chinese, are used with different sets of counters.
Typically the core numerals are more “free” than their Chinese counterparts, e.g. in replacing
A Reference Grammar of Korean
PART I 175
constructions of numeral + counter. But there are no core numerals for hundred, thousand, or ten
thousand in modern Korean, so where we would expect a core morpheme the Chinese numeral is used
instead:
99 people
100 people
101 people
199 people
20,002 people
ahun ahop salam
payk salam
payk han salam
payk ahun ahop salam
i-man twu salam
kwu-sip kwu-myeng
payk-myeng
payk il-myeng
payk kwu-sip kwu-myeng/-in
i-man i-myeng
The Chinese morpheme pan ‘half ••• ; ••• and a half’ is also used where the core set is appropriate: pan
sikan ‘half an hour’, twu sikan pan ‘two hours and a half’, twu si pan ‘2:30 o’clock’.
Some of the core numerals have shortened shapes when they are in modifying position: chayk twu
kwen but chayk twOl ‘two books’. The full shape, however, usually appears before the postnoun (the
“postnumeral”) ccay, so that ordinarily in Seoul ‘second’ is twul ccay, though less commonly (in
dialects) you will hear the shortened shape: twu ccay = twul ccay. There seems to be confusion over
whether to use the shortened shape of a numeral before ccay. The full shape is more common for
‘second’ (twul ccay), as we have just said, and the shortened shape is used for ‘eleventh’ (yel han ccay
- similarly for ‘21st’, ‘31st’, ...) and for ‘twentieth’ (sumu ccay), but the longer forms are also found:
yel hana ccay and sumul ccay. For ‘third’ and ‘fourth’ it is purely a spelling problem, since the
pronunciation would be identical in either case, owing to the way the morphophonemic rules work.
The prevailing spelling standard in South Korea favors the full forms: seys ccay rather than sey ccay,
and neys ccay rather than ney ccay. But the North Korean grammar CM prefers the short forms. And
Ceng Insung 1960:190-2 tries to set up a distinction between (1) ches ccay, twu ccay, sey ccay, ... ,
yel han ccay, ... , sumu ccay and (2) hana ccay, twul ccay, seys ccay, ... , yel hana ccay, ... , sumu(l)
ccay. In the meaning equivalent to - pen ccay ‘-th’ he would use the first group, the shortened forms,
and the second group would be used when ••• ccay is a synonym of - chay ‘and all, the whole, intact’.
But my informants say that it is awkward to make combinations of numeral + chay (or the ccay that is
a synonym of it). Instead, they prefer to insert a counter: yel han kay ccay means either ‘the eleventh
(thing)’ or (= yel han kay chay) ‘all eleven (things)’. Yet you may run across hana ccay/chay in the
meaning han kay ccay/chay ‘one whole (thing)’.
This postnumeral element ccay makes the expected ordinals for all non-fractional core numerals,
but where *han(a) ccay would be expected we find instead a unique compound of adnoun + postnoun:
ches ccay ‘first’. However, the string han ccay will turn up in yel han ccay ‘11th’, sumul han ccay
‘21st’, payk han ccay ‘101st’, ... . We also find payk ccay ‘100th’, chen ccay ‘1,000th’, ... . Moreover
ccay occurs also with meych (‘how-manyeth’) and with some of the approximate numerals (Cf Choy
Hyenpay 1959:566).
The Chinese numerals are made ordinal by the adnoun cey ••• ‘-th’: cey-il ‘first’, cey-i ‘second’,
cey-sam ‘third’.And we find cey payk (etc.) as well as payk ccay for ‘100th’; notice that
‘hundred and first’ is either cey payk-il or payk han ccay.
The core numerals for ‘3’ and ‘4’ have the special shapes sek and nek before certain counters
(usually beginning with t- or c-) and se or ne before certain others. For some of the counters there is
variation between the several alternants - as, for some, there is a choice between using Chinese or
core numerals. The numerals yel ‘ten’ and yetel(p) ‘eight’ are treated as yelq and yetelq before
counters that begin with a plain obstruent (p t c s k): yelq kay ‘10 things’, yelq tay ‘10 machines (or
vehicles)’, yelq cang ‘10 sheets’, yelq pen ‘10 times’ (Cf il-pen ‘number one’), yelq pun ‘10 people’
(Cf phal-pun ‘eight minutes’), yetel[p]q pam ‘eight nights’, yetel[p]q si ‘8 o’clock’. The shape yelq
also appears in yelq-twul/-twu ‘twelve’.
Many of the odd forms of the numerals are regularized by younger speakers; sometimes there is a
difference of meaning or nuance. Or the irregular form is heard in set phrases: a person who says sey
tal(q tong-an) ‘three months (long)’ - and, being young, counts ten days as sip-il - may nonetheless
176 PART I
A Reference Grammar of Korean
say sek-tal yel.hul ‘3 months and 10 days = 100 days’ but only because that is a lexicalized phrase
with special significance.
The following lists are designed for convenient reference. Theoretical problems of inclusion,
arrangement, and the like, are passed over in silence. The morphemic structure of the core numerals
involves various alternations of shape; the alternants are shown in the right column of the first list as
“bound core elements”.
(1) List of cardinal numerals
Chinese numerals
0 *yeng, kong
Core numerals
Bound core elements
V 2 pan, celpan
••• (k)awus 1
••• pan
1 il
2 I; *yang ‘both’
3 sam
4 sa
5 6
6 ’yiik
7 chil
8 phal
9 kwu
10 sip
11 sip-il
12 sip-i
13 sip-sam
14 sip-sa
15 sip-o
16 sip-lyuk (sipq-’yuk)
17 sip-chil
18 sip-phal
19 sip-kwu
20 I-sip
21 i-sip il
22 I-sip I
23 i-sip sam
24 i-sip sa
25 i-sip 6
26 I-sip(q) J yuk*
27 i-sip chil
28 i-sip phal
29 i-sip kwu
30 sam-sip
33 sam-sip sam
40 sa-sip
44 sa-sip sa
50 o-sip
55 o-sip 6
60 ^uk-sip
66 'yuk-sipq *yuk
70 chilq-sip
77 chilq-sip chil
hana / han
twfll / twu
seys / sey/sek/se
neys / ney / nek / ne
tases
yeses/yes 3
ilkop
yetel(p) / yetel[p]q 5
ahop
yel/yelq 6
yel-hana/-han
yelq-twul/-twu
yelq-seys / -sey / -sek / -se
yel-neys / -ney / -nek / -ne
yelq-tases
yel(q)-yeses
yel(q)-ilkop
yel(q)-yetel(p)
yel-ahop
sumul / sumu
sumul hana / han
sumulq twfll / twu
sumulq seys / sey / sek / se
sumulq neys / ney / nek / ne
sumulq tases
sumul(q) yeses
sumul(q)
sumul(q) yetel(p)
sumul ahop
sel(h)un 7
sel(h)un seys / sey / sek / se
mahun
mahun neys / ney / nek / ne
swln, [dialect] swihun
swin tases
yeyswun
yeyswunq yeses
ilhun
ilhunq ilkop
hanak 2 , ha
it
sen, sa, sel
net, na, ma
tays, tay, tas, ta, swl
yes, yeys, yey
il, nil, nilkop 4
yetul, yet
ahu
(un, hun, n, wun)
*/Isimnyuk, isiplyuk/
A Reference Grammar of Korean
PART I 177
80 phalq-sip
yetun, [dialect] yatun
88 phalq-sip phal
yetunq yetel(p) / yetel[p]q
90 kwu-sip
ahun (but ahu before -nam.un)
99 kwu-sip kwu
ahun ahop
100 payk, (il-payk)
- (obsolete on < ’won)
101 payk il
payk hana /han
115 payk sip-o
payk yelq-tases
144 payk sa-sip sa
payk mahun neys/ney/nek/ne
200 I-payk
300 sam-payk
306 sam-payk(q) 'yuk*
*/sampayngnyuk/ or
400 sa-payk
500 o-payk
600 *yuk-payk
700 chil-payk
800 phal-payk
900 kwu-payk
/sampayk 1 yuk/
1000 chen (il-chen) 1 2 3 4 5 6 7 8
2000 i-chen
3000 sam-chen
4000 sa-chen
5000 o-chen
6000 'yuk-chen
7000 chil-chen
8000 phal-chen
9000 kwu-chen
- (obsolete cumun < cu muti)
10 000 man (i 1-man)
- (? kol, kkol) 9
100 000 sip man
- (? cal)
1 000 000 payk man
100 000 000 ek (il-ek)
1 million million co (ilq-co)
— (? wul)
1 Fairly limited: sek ca kawus ‘three and a half ca’, twu mal kawus ‘two and a
half mal’, toy kawus ‘one and a half toy’. As the last example shows, han - is
usually not expressed when kawus is added.
2 Hanak is an occasional free variant of hana before ssik.
3 The shape yes occurs before nyang, toy, mal, pal.
4 The shape is il in iley ‘7 days’, nil in yey-niley ‘6 or 7 days’, and nilkop in yey-
nilkop ‘6 or 7’.
5 Dialect yatal, yatul, yetup. In Seoul (and widely) the ™p surfaces only in yetelp
hay ‘eight years’; in dialects it will also be heard in yetelp-i ‘eight people’ (not
currently used in Seoul).
6 But ye - before -nam.un.
7 The form without the h is preferred.
8 The version il-chen is used only in arithmetic or meticulous listing. If a counter
follows, il- is not used: senswu chen-myeng ‘a thousand athletes’.
9 See CM 1:307. Is this (as suggested by Sin Kichel 1958:117), based only on the
set expression kol payk pen ‘many many times’ and the synonymous (k)kol chen
pen? There are few (if any) examples of the last three numerals, which are said to
be archaic.
178
PART I
A Reference Grammar of Korean
(2) List of ordinal numerals
Chinese
1st
cey-il
2nd
cey-i
3rd
cey-sam
4th
cey-sa
5th
cey-o
6th
cey-'yuk (/ceyyuk/ only!)
7th
cey-chil
8th
cey-phal
9th
cey-kwu
10th
cey-sip
11th
cey sip-il
12th
cey sip-i
20th
cey I-sip
100th
cey-payk
133rd
cey-payk sam-sip sam
how-manyeth
-
(3) List of approximate cardinal numerals
1-2
Chinese
il(q)-i /il-I, il-lT/
2
2-3
1-sam
3
3-4
sam-sa
4
4-5
sa-o
5
5-6
o^yuk /o.yuk, olyuk, onyuk/
6-7
‘yuk-chil
7
7-8
chil-phal
8-9
phal-kwu
9-10
—
10 +
sip-ye
10-20
il(q)-i sip
20 +
Isip-ye
20-30
I-sam sip
30 +
samsip-ye
30-40
sam-sa sip
40 +
sasip-ye
Core
ches ccay; uttum (‘top’); [dialect] han ccay
twul ccay (less commonly twO ccay;
“it-ccay, is-ccay” seem to be artificial)
seys ccay (also spelled sey ccay)
neys ccay (also spelled ney ccay)
tases ccay
yeses ccay
ilkop ccay
yetel(p) ccay
ahop ccay
yel ccay
yel-han ccay
yelq-twul ccay
sumu ccay (less commonly sumul ccay)
payk ccay
payk sel(h)un seys ccay
meych ccay < myech ccay
Core
han(a)-twul, han-twu
han-twues/-twue ‘1 or 2’
twues/twue ‘about 2’
twu(l)-seys / -sey / -sek / -se
twu-senes /-sene ‘about 2 or 3’
senes/sene ‘about 3’
sene-netes /-nete ‘about 3 or 4’
netes / nete ‘about 4’
nete-tays ‘about 4 or 5’
tays ‘about 5’
tay-yeses ‘about 5 or 6’
ye(y)-nilkop ‘about 6 or 7’
il(ko)-yetel(p) ‘about 7 or 8’
yet-ahop ‘about 8 or 9’
yeles/yele ‘about 10; several, many’
ye-nam(.)un/-nam(.)u ‘10-odd’
[dialect yelamu(n)]
sumu-nam(.)un/-nam(.)u ‘20-odd’
sel(h)un-nam(.)un / -nam(.)u,
sel(h)un namcis ‘30-odd’
mahun-nam(.)un / -nam(.)u,
mahun namcis ‘40-odd’
A Reference Grammar of Korean
PART I 179
90 +
kwusip-ye
ahu-nam(.)un / -nam(.)u,
ahun namcis ‘90-odd’
90-100
-
—
100 +
payk-ye
payk namcis ‘100-odd’
110 +
payk sip-ye
-
120 +
payk Isip-ye
-
130 +
payk samsip-ye
-
Notice also: swu-sip ‘several tens (of)’, swu-payk ‘several hundred’, swu-chen ‘several thousand’,
swu-man ‘tens of thousands (of)’, swu-ek ‘hundreds of millions (of)’.
(4) List of approximate ordinal numerals
Chinese
Core
2nd or so
-
twue ccay
2nd or 3rd
cey I-sam
twu-sey ccay
twu-sene ccay ‘about 2nd or 3rd’
3rd or so
-
-
3rd or 4th
cey sam-sa
sene-nete ccay ‘about 3rd or 4th’
Other such forms seem awkward, especially ches-twu ccay and yele ccay. But yele(s) ccay can be used
as an abbreviation of yele pen ccay, as in the following exchange: I pen ey nah.un Kim ssi ney ai ka
ches ayki ’n ka yo? - Kulssey olssita; ama yele(s) ccay toylq ke 1’ yo ‘Is this the first child for the
Kims? - I don’t think so, it must be one of several’.
(5) List of excess ordinal numerals
Chinese
10th or so sip-ye ccay
20th or so Isip-ye ccay
Core
ye-nam(.)u ccay
sumu-nam(.)u ccay
90th or so
100th or so
1,000th or so
10,000th or so
kwusip-ye ccay
payk-ye ccay
chen-ye ccay
man-ye ccay
ahu-nam(.)u ccay
5.5.3. Counters.
The following list of counters is not quite exhaustive, but it is fairly representative. There are
three columns: the first lists the typical units counted, the second shows an example with core
numerals, the third shows an example with Chinese numerals. In general, the examples are given with
the numeral ‘3’ in order to show which counters take the shapes sek or se (and for ‘4’ nek or ne). If
there is a blank in the core or Chinese column, the counter does not normally occur with those
numerals. However, the Chinese numeral must be used when there is no core numeral (‘100’, ...).
List of counters: group one
Units counted with core numerals with Chinese numerals
things, items,
matches, pencils (mulken sey) kay ‘3 objects’
units, items, bits,
grains, beans (khong sey) nath ‘3 beans’
books, magazines;
20 sheets of Korean paper (chayk sey) kwenj ‘3 books’
animals, birds, fish (kay sey) mali ‘3 dogs’
180 part i A Reference Grammar of Korean
animals (horse, ox)
honored persons
persons, people
persons [formal]
flat things, sheets, papers,
newspapers, letters
sheets, mats
buildings
vehicles, machines
long objects with handles
(bushes, brooms, guns,
scythes); pencils (but
kay is more common)
cigarettes; pipefuls
slaps
suits of clothes, garments;
sets of dishes / tableware;
copies of a set of documents
ten garments/dishes
(menu) dishes
pairs of footwear or gloves
cannons, big guns
trees, shrubs
places, institutions
places, locations, spots
fields
boats
small round things (berries,
nuts, beads, bullets, lenses)
poems
written characters, letters
chapters (of text)
pieces of sewing thread
skeins of thread
pairs of chopsticks
hung pictures (any kind); scrolls
agenda items, assembly bills
kinds, sorts
(so sey) phil ‘3 oxen’
(son nim sey) pun ‘3 guests’
(salam sey) salam ‘3 people’
(haksayng sey) myeng ‘3 students’
(sinmun sek) cang ‘3 newspapers’
(cali sey) ttwayki ‘3 mats’
(cip sey) chay ‘3 houses’
(catong-cha sek) tayj
‘3 automobiles’
(yangpok sey) pel ‘3 suits’
(cekoli / kulus sek) cwuk
‘30 vests/plates’
(Cwungkwukq Voli sey) cepsi
‘3 Chinese dishes’
(sin-pal sey) khyel(l)ey
‘3 pairs of shoes’
(namu sey) kulwu ‘3 trees’
(thullin tey sey) kwuntey ‘3 errors’
(non sey) paymi ‘3 fields’
(pay sey) chek / chay ‘3 boats’
(photo / pam sey) al
‘3 grapes /chestnuts’
(si sey) swu ‘3 poems’
(kulqca sek) ca ‘3 letters’
(kul sek) cang ‘3 chapters’
(sll sey) nim ‘3 pieces of thread’
(sil sey) they ‘3 skeins of thread’
(ceq-kal sey) may ‘3 pairs of
chopsticks’
More commonly just ceq-kal seys.
(kulim sey) phok ‘3 pictures’
(sey) ken ‘3 items/bills’
(os sey) kaci ‘3 (kinds of) garments’
(kwunma sam)-phil
‘3 army horses’
(nongwu sam)-twu ‘3 farm
oxen’
(cwuthayk sam)-ho / -tongj
‘3 dwellings’
(catong-cha sam)-tay
‘3 automobiles’
(taypho sam)-mun ‘3 guns’
(namu sam)-cwu ‘3 trees’
(kongcang sam)-kayso
‘3 factories’
(si sam)-swu ‘3 poems’
(sam)-ken ‘3 items/bills’
(pus sek) calwu ‘3 writing brushes’
(tampay sey/sek) tay 2 ‘3 cigarettes’
(ppyam sek) tay 3 ‘3 slaps (on cheek)’
A Reference Grammar of Korean
PART I 181
Units counted
List of counters: group two
with core numerals
bunches (of vegetables/firewood)
bunches (of flowers, plants)
bunches, clumps (of false hair,
seaweed)
sheaves, tied bunches;
strings of tobacco
sheaves of straw
bundles of chopped firewood
loaves
pinches, dashes (of spice/herbs)
fist(ful)s, handfuls
mouthfuls, sips
bottle(fuls)
cup(ful)s
bowl(ful)s
box(fuls), small
box(fuls), large
cratefuls
packet(ful)s
packs (of herbal remedies)
20 packs (of herbal remedies)
bag(fuls)
cakes, blocks, squares
(human) backloads
(pack)loads
loads, bundles, packs,
pieces of luggage
bolts of cloth
bolts of cloth; heads
of cabbage, gourds
letters, telegrams
24 needles
100 raincovers
(namul sek) tan ‘3 bunches of greens’
(kkoch sey/sek) tapal ‘3 bouquets’
(myek sey) kkokci ‘3 clumps of seaweed’
(pye sek) cwul ‘3 sheaves of rice’
(ciph sey) muSi ‘3 sheaves of straw’
(cangcak sey) mus 2 ‘3 bundles of firewood’
(ppang sey) tengeli ‘3 loaves of bread’
(yangnyem sey) capam ‘3 dashes of spice’
(molay sey) moswum ‘3 handfuls of sand’
(mul sey) mokum ‘3 sips of water’
(maykcwu sey) pyeng ‘3 bottles of beer’
(cha sek) can ‘3 cups of tea’
(pap sey) kulus ‘3 bowls of rice’
(sengnyang sey) kap ‘3 boxes of matches’
(kwaca sey) sangca ‘3 boxes of cakes’
(sakwa sey) kwey-ccak ‘3 crates of apples’
(yak sey) pong ‘3 packets of medicine’
(yak sey//sek) chep ‘3 packs of herbs’
(yak sey/sek) cey ‘60 packs of herbs’
(ssal sey) kama} ‘3 bags of rice’
(twupu sey) mo ‘3 cakes (squares) of bean curd’
(namu sey) cim ‘3 (back)loads of wood’
(koksik sey) pali ‘3 (pack)loads of grain’
(cim sey) ccak ‘3 pieces of luggage’
(philyuk sey) phil ‘3 bolts of cloth’
(kwangmok / paychwu sey) thong
‘3 bolts of cloth/cabbages’
(cenhwa sey) thonghwa ‘3 calls’
(panul sey) ssam ‘3 dozen needles’
or tobacco-pouches (kalmo/ssamci sey) kama ‘300 raincovers/pouches’
tied bundles of 10
flat dry edibles (kwulpi sey) kas ‘3 bundles of dried corvina’
10 eggs in a straw wrapper (talkyal sey) kkwule(y)mi ‘3 wrappers of eggs’
bundles of 50 cucumbers or
eggplants (oi/kaci sey) keli ‘150 cucumbers / eggplants’
reams (500 sheets) of paper (congi sey) *yen ‘3 reams of paper’
20 sheets of Korean paper (congi sey) kwen 2 ‘60 sheets of paper’
200 sheets of Korean paper;
a roll of paper; 20 almanacs (congi sey) chwuk ‘600 sheets of paper’
100 fruits, radishes, cabbages,
bulbs of garlic (kam sek) cep ‘300 persimmons’
plants, heads (of cabbage) (paychwu sey) phoki ‘3 heads of cabbage’
182 PART I
A Reference Grammar of Korean
a bundle (of 10 brushes pus,
50 bolts of cloth pey,
200 herring piwus)
fish (as a commodity)
handfuls of fish
(2 large, 4-5 small fish)
10 fish
20 fish
20 cuttlefish
20 pollacks [rare]
2000 fish
bunches of barley
30 bunches of barley
1500 bunches of barley
2000 tiles
Units counted
(hours) o’clock
hours
nights
weeks
months (See separate list.)
years
years old
spells, periods (of activity)
seconds
minutes
parts, fractions
ten-percents
degrees
times
moves (in chess / checkers)
(gun)shots
rounds (esp. of 5 arrows shot)
floors, stories
(...th) floor / story
wen, yen, yuan, dollar
hwan [obsolete = wen]
cen, sen, cents
(old copper coin)
(old coin) taels
(old coin) yepcen
(... sek) tong 2 ‘30 (brushes), 150 (bolts),
600 (herring)’
(mulq-koki sey) kay ‘3 fish’
(kotunge/coki sey) son ‘6 mackerels/corvinas’
(mulq-koki sey) mus ‘30 fish’
(mulq-koki sek) tulem ‘60 fish’
(ocinge sey) chwuk ‘60 cuttlefish’
(myengthay / puk.e sey) khway ‘60 pollacks /
dried pollacks’
(mulq-koki sey) pali ‘6000 fish’
(poli sek) tan ‘3 bunches of barley’
(poli sey) had ‘90 bunches of barley’
(poli sek) tong 3 ‘4500 bunches of barley’
(kiwa sey) wuli ‘6000 tiles’
List of counters: group three
with core numerals with Chinese numerals
(sey) si ‘3 o’clock’ 1
(sey) sikan ‘3 hours’
(sey) pam ‘3 nights’
(sey) cwukan/cwuil ‘3 weeks’
(sek) tal ‘3 months’
(sey) hay ‘3 years’
(sey) sal ‘3 years old’
(sey) cham ‘3 spells’
(sey) pen ‘3 times’
(sey) swu ‘3 moves’
(sey) pang ‘3 shots’
(sey) swun ‘3 rounds’
? (sey) chung ‘3 floors’
? (sey) chung ‘3rd floor’
(sam)-cwukan / -cwuil
(sam)-kaywel ‘3 months’
(sam)-nyen ‘3 years’
(sam)-sey ‘3 years old’
(sam)-cho ‘3 seconds’
(sam)-pun ‘3 minutes’
(sam)-pun ‘3 parts’
(sam)-hal ‘30 percent’
(Sep-ssi sam)-to
‘3° centigrade’
(wito/kyengto sam)-to
‘3° longitude /latitude’
(sam)-hoy ‘3 times’
(sam)-pal ‘3 shots’
(sam)-chung ‘3rd floor’
(sam)-wen ‘3 wen’
(sam)-hwan ‘3 hwan’
(sam)-cen ‘3 cen’
(sey) niph ‘3 coppers’
(sek) nyang ‘3 taels’
(yepcen sam)-mun ‘3 yepcen’
A Reference Grammar of Korean
part i 183
(old dime = 10 phun)
(se) ton ‘3 old dimes’
(old Korean penny)
(se) phun ‘3 pennies’
dollars
(sam)-pul ‘3 dollars’
marks
(sam)-malukhu ‘3 marks’
rubles
(sam)-lwupul ‘3 rubles’
pounds
(sam)-phauntu /-pang ‘3 pounds’
shillings
(sam)-silling ‘3 shillings’
liras
(sam)-lila ‘3 lira’
francs
(sam)-phulang ‘3 francs’
1 Also (sek) cem, said to be used by “
‘the uneducated”.
List of counters: group four
Units counted
with core numerals
with Chinese numerals
(1) Linear measure
.0303 mm
(sam)-mo
.303 mm
(sam)-li
3.03 mm = .119 in
(sek/se) phun
(sam)-phun
3.03 cm = 1.193 in
(sey) chi
(sam)-chon
.303 m = 0.994 ft
(sey/sek) ca 1
(sam)-chek
1.818 m = 1.988 yd
(sey) kan
(sam)-kan
109 m = .542 furlongs
(sam)-ceng
3.927 km = 2.44 m
(sam)-ll 2 3
1 Yeses ca ‘6 ca’ is often said as ye(y) ca.
2 But this counter
is generally used only in multiples
of five. Traditionally
it refers to several different lengths; the best overall translation is the equally
vague English “leagues”.
(2) Square measure
330.7 sq cm = .355 sq ft (sam)-cak
33.07 sq cm = 3.556 sq ft (sey/se) hop
3.307 sq m = 3.952 sq yd (sey) phyeng
91.15 sq m = 3.92 sq rods (sam)-mu
9.915 ares = .245 acres (sam)-tan
(3) Liquid and dry measures
.018 liters = .152 gills (sam)-cak
.18 liters = 1.524 gills (sey/se) hop
1.805 liters = 3.81 pints
(sek/se) toy
(sam)-sung [rare]
18.05 liters = 19.04 qt
(se) mal
(sam)-twu [rare]
180.5 liters = 47.6 gal
(sek) sem
(sam)-sek; (sam)-kok [rare]
(4) Weight measures
.003759 gram
(sam)-mo
.03759 gram
(sam)-li
.3759 grams
(se) phun
3.759 grams = 2.117 drams
= .13228 ounces
(se) ton
.601 kg = 1.323 lb
(sey/se) kun
3.759 kg = 8.27 lb
(sey) kwan
1.803 kg = 3.969 lb = 30 kun
(sam)-kyun 1
1 Rare, except in chen-kyun pota mukepta ‘is ever so heavy’.
184 PART I
A Reference Grammar of Korean
The interrelationships of the measure units can be displayed as follows:
(1)
10 mo =
1 li
(3)
10 cak =
1 hop
10 li
1 phun
10 hop =
1 toy
10 phun =
1 chi
10 toy =
1 mal
10 chi =
1 ca
10 mal =
1 sem
6 ca =
1 kan
10 kan =
1 cang
6 cang =
1 ceng
36 ceng =
1 IT
(2)
10 cak =
1 hop
(4)
10 mo =
1 li
10 hop =
1 phyeng
10 li
1 phun
30 phyeng =
1 mu
10 phun =
1 ton
10 mu =
1 tan
160 ton =
1 kun
10 tan =
1 ceng
1000 ton =
1 kwan
Units counted
fathoms (8 or 10 ca; 5 ca)
grams
kilo(gram/watt/meter)s
spans
(double-)span of rope;
10 handspans of rope
handspans of rope
50 pal of rope
spans (of thread/string/rope)
double armspans (around)
inches
feet
yards
miles
“miles, leagues” (? = IT,
when less than ten)
ounces
pounds
tons
generations, the •th
sizes of rice field
(in terms of the yield)
meals
wins (at wrestling)
times (as much)
(dawn) cock-crows
List of counters: group five
with core numerals
(sey) kil
(sey) kulam
(sey) khillo
(se) pal
(sey) macang
(sey) thon
(sey) ma’-ciki;
(sey) toy-ciki
(sey) kki
(sey) heli
(sey) kopcel 1 ‘3 times as much’
(sey) hway (ccay) ‘3 cock-crows
(3rd cock-crow)’
with Chinese numerals
(sam)-cang
(sam)-aunsu
(sam)-phauntu
(sam)-thon/-ton
(Heyn.li sam)-sey ‘Henry III’
(saykki sey) pal ‘3 pal of rope’
(saykki sey) ppyem
‘3 handspans of rope’
(saykki sek) tong 4 ‘150 pal of rope’
(sll/no-kkun/saykki sey) palam
(sey) alum
(sey) inchi
(sey) phlthu
(sey) yatu, ma
(sam)-mail
shoe-sizes
(sam-)mun ‘size 3’
A Reference Grammar of Korean
PART I 185
dozens (sey / sek) tha
(male-female) pairs (say sey) ssang ‘3 pairs of birds’
(thread weave density) (mumyeng sek) say
‘3-thread cotton cloth’
one of a pair (sin-pal sey) ccak
(Cf han-ccak ‘a set’) ‘3 odd (unmatched) shoes’
1 The word kapcel usually means only ‘two-fold’; ‘double’ is kop or kop-cayngi.
5.5.4. Irregular counting.
A few units are counted in irregular ways: days, years, months; people; cattle and horses (of
certain ages).
(1) DAY; DAY OF MONTH
Normal
Formal
how many days; what day
meych nal
meychil (myechil);
(of the month)
meychit nal
1 day; 1st of month
halwu; halwuq nal
il-il
2 days; 2nd of month
ithul; ithut nal
i-il
3 days; 3rd of month
sahul; sahut nal
sam-il
3 or 4 days
sanal
sam-sa-il
4 days; 4th of month
nahul; nahut nal
sa-il
4 or 5 days
natal
sa-o-il
5 days; 5th of month
tassa; tassayq nal
o-il
6 days; 6th of month
yessay; yessayq nal
1 yuk-il
6 or 7 days
yey-niley
1 yuk-chil-il
7 days; 7th of month
iley; ileyq nal
chil-il
8 days; 8th of month
yetuley, [dialect] yatuley;
yetuleyq nal
phal-il 2
9 days; 9th of month
ahuley; ahuleyq nal
kwu-il
10 days
10th of month
yelhul, yelhul nal
sip-il
11 days
11th of month
yel halwu; yel halwuq nal
sip-il il
12 days
12th of month
yel ithul; yel ithut nal
sip-i il
13 days; 13th of month
yelq sahul; yelq sahut nal
sip-sam il
14 days; 14th of month
yel nahul; yel nahut nal
sip-sa il
15 days; 15th of month
yelq tassay; yelq tassayq nal
sip-o il
the midmonth (day = the 15th)
polum (nal)
16 days
16th of month
yel yessay; yel yessayq nal
sipq- ! yuk il
17 days
17th of month
yel iley; yel ileyq nal
sip-chil il
18 days
18th of month
yel yetuley; yel yetuleyq nal
sip-phal il
19 days
19th of month
yel ahuley; yel ahuleyq nal
sip-kwu il
20 days
20th of month
sumu nal
i-sip il
21 days
21st of month
sumul halwu
i-sip il-il
22 days
22nd of month
sumu ithul
I-sip i-il
23 days
23rd of month
sumu sahul
i-sip sam-il
24 days
24th of month
sumu nahul
i-sip sa-il
25 days
25th of month
sumulq tassay
i-sip o-il
26 days
26th of month
sumul yessay
i-sip(q) 'yuk-il
27 days
27th of month
sumu iley
i-sip chil-il
28 days
28th of month
sumu yetuley
i-sip phal-il
29 days
29th of month
sumu ahuley
I-sip kwu-il
30 days
30th of month
(? selun nal)
sam-sip il
31 days
31st of month
(? selun halwu)
sam-sip sip-il
the last day of the month
kumum (nal)
186 PART I
A Reference Grammar of Korean
From 1-31, the terms either count days or name the days of the month; but, unless it is the only
form, the term with ••• nal usually just names. For phal-il the variant pha’-il means only the 8th of the
month, and it is usually taken to refer to the 8th of April, Buddha’s birthday.
For 21-31 the native forms are uncommon. They are usually replaced by the Chinese forms, and
that may account for the unexplained choice by my sources of sumu or sumul in a given expression. In
dialects more comfortable with the older forms the choice may be better motivated.
When the first ten days of the month are designated it is customary to attach the adnoun cho(-): cho
halwu, cho ithul, ... , cho yelhul.
To designate a quantified period of time -q tong-an is often added: yelhulq tong-an ‘10 days’ =
sip-ilq tong-an ‘ten days (duration)’, polumq tong-an = sip-o il(q) tong-an ‘fifteen days’. The
postnoun (or suffix) kan ‘interval’ is often added to the terms for ‘21-31 days (duration)’ (Cf M
1:173): sam-sip il-kan(q tong-an) ‘(a period of) 31 days’.
Naming days of the week
what day
musun nal, musunq yoil
Sunday
il.yo(il (nal))
Monday
wel.yo(il (nal))
Tuesday
hwa.yo(il (nal)
Wednesday
swuyo(il (nal))
Thursday
mok.yo(il (nal))
Friday
kum.yo(il (nal))
Saturday
tho.yo(il (nal))
(2) Years
how many years
meych hay
meych nyen
what (which) year
musun/enu hay (of 60-yr cycle,...)
meych nyen (of calendar)!
1 year
han hay
il-nyen (also Year 1)
2 years
twu hay, ithay
i-nyen (also Year 2)
3 years
sey hay
sam-nyen (also Year 3)
4 years
ney hay
sa-nyen (also Year 4)
5 years
tases hay
o-nyen (also Year 5)
6 years
yeses hay
’yuk-nyen (also Year 6)
7 years
ilkop hay
chil-nyen (also Year 7)
8 years
yetelp hay /yetelphay/
phal-nyen (also Year 8)
9 years
ahop hay
kwu-nyen (also Year 9)
10 years
yel hay
sip-nyen (also Year 10)
20 years
sumu hay 2
l-sip nyen (also Year 20)
100 years
payk hay
payk-nyen (also Year 100)
1 Seki 1992 = Tanki 4325. (The myth says Korea began in 2333 B.C.)
2 But sumul hoy (1481 Twusi 16:18a).
(3) Months
how many months
meych tal
meych-kaywel
1 month
han tal
il-kaywel
2 months
twu tal
I-kaywel
3 months
sek tal
sam-kaywel
4 months
nek tal
sa-kaywel
8 months
yetel[p]q tal
phal-kaywel
10 months
yelq tal
sip-kaywel
A Reference Grammar of Korean
PARTI
187
what month
musun tal
meych wel, musun wel
January
il-wel(q tal); ceng-wel
February
I-wel(q tal)
March
sam-wel(q tal)
April
sa-wel(q tal)
May
o-wel(q tal)
June
‘yu’-we^q tal)
July
chil-wel(q tal)
August
phal-wel(q tal)
September
kwu-wel(q tal)
October
si’-wel(q tal)
November
sip.il-wel(q tal); tongciq tal
December
sip.i-wel(q tal); set tal
(4) People
In Seoul, people are usually counted regularly with han salam, twu salam, ... , or (more formal)
han myeng, twu myeng, ... , and in compounds and set expressions also il-in, I-in, ... , as in il-in (i-in)
yong pang ‘a room for one (two)’. In certain other areas, people are counted with the bound noun i
‘one, person’ as follows: 1 person han(a)-i, hanq-i /hanni/
2 people twOl-i, twu-i
3 people se-i
4 people ne-i
5 people tases-i, tasesq-i
6 people yeses-i, yesesq-i
7 people ilkop-i, ilkopq-i
8 people yetel-i, yetelp-i
9 people ahop-i, ahopq-i
10 people yel-i, yelq-i
20 people sumul-i
Ko Yengkun (1989 LR 25:102) gives the forms myech-i ‘how many people’ and yeles-i ‘many people’
but I have been unable to confirm them. Perhaps there is confusion with myech i = meych i and yeles
i, in which i is the nominative marker and there is nothing referring to people, as such. Choy Hak.kun
(1978:1048) lists the form yelesi as a dialect equivalent to yeles and not specifically meaning people;
this is an example of the common incorporation of -i by nouns that is described in Part II.
(5) Horses and oxen of a certain age
1-year-old
hansup
2-year-old
twOsup
3-year-old
salup; [dialect] seysup
4-year-old
nalup
5-year-old
tasup
6-year-old
yesup
7-year-old
ilop
8-year-old
yetup
9-year-old
asup, kwulup
10-year-old
yellup, tam(p)ul
There are also a few variant forms of the numerals with certain of the counters: tay ca, ta(s) ca =
tases ca (Cf tays ca ‘about 5 feet’), yes ca = yeses ca; tas ton = tases ton, yes ton = yeses ton.
In the game of tag (swullay capki) the counting goes: 1 hanalttay, 2 twualttay, 3 samacwung,
4 nalttay, 5 - (?), 6 ^uk-nangkeci, 7 - (?), 8 phalttay, 9 cangkwun, 10 kotulay-ppyong.
188 PART I
A Reference Grammar of Korean
5.5.5. Fractions.
Examples of numeral fractions and how they are read:
1/2 I-pun uy/ci il; celpan
V 3 sam-pun uy/ci il
2 /3 sam-pun uy/ci I
3 /4 sa-pun uy/ci sam
5 / 6 ‘yuk-pun uy/ci 0
7 /g phai-pun uy/ci chil
1.3 ilq-cem sam (“one-point three”)
2.1 I-cem il
(0).314 ^yengq-jcem sam il sa
Some grammarians treat (-)punci as a unit (suffix or postnoun); the ci is a Chinese particle equivalent
to the core adnominal marker uy.
5.6. Verbal nouns.
A verbal noun is typically followed by a postnominal verb. Many verbal nouns are pre-
separable, in that they are followed by separable postnominal verbs such as ha- ‘do/be’, toy- ‘get
done’, sikhi- ‘cause to do’, ka- ‘go to do’, (na-)o- ‘come (out) to do’, po- ‘see to it, do’, ... . These
verbs are called separable because they are sometimes separated from certain verbal nouns by the
insertion of a particle - at least tul, to, man, or un/nun; and often ul/lul or i/ka. Those verbal
nouns from which a separable postnominal can be separated are called separable verbal nouns; Cf
H ankul 108:42 (1955). Most of the two-syllable verbal nouns are separable and most of the one-
syllable verbal nouns are inseparable, but there are exceptions. Some of the verbal nouns are pre-
inseparable: they occur before inseparable postnominal verbs such as ha- ‘behave, go (boom!, ... )’
(a homophone of ha- ‘do/be’ and the same etymon), keli- ‘behave repeatedly’, tay- = keli-, k(h)uli-
‘behave’, sulew- ‘be, give the impression of being’, ... . These verbs are called inseparable because
they do not allow a particle (not even to or un/nun) to intervene. (But certain speakers allow sulepta
to be set apart in vivid contexts: Calang to sulewe haney! ‘How proud he is!’; saiang/iksai/yatan to
sulepta ‘is quite lovely/droll/irksome’, ... .) The inseparable postnominal verbs are sometimes
attached to pre-inseparable verbal nouns, sometimes to separable verbal nouns, i.e. to verbal nouns that
can occur with particles when in construction with separable postnominal verbs.
Some verbal nouns occur only in constructions of verbal noun (with or without particle) +
postnominal verb. (But note that the postnominal verb hanta is sometimes dropped, especially when
the forms ham and hako are used in headlines; see §10.2, llb-d.) Other verbal nouns occur also as
free nouns, for example kongpu ‘study’ in i kongpu ka elyepta ‘this study is difficult’. We are dealing
with three independent variables of the grammar:
(1) verbal noun only # free noun also;
(2) pre-separable verbal noun + pre-inseparable verbal noun;
(3) separable verbal noun £ inseparable verbal noun.
For each verbal noun in the lexicon, such three-way information should be sought. (See below for two
more pieces of information that we require.) Most Korean dictionaries list as free nouns certain items
that seem to be limited, in speech at least, to use as verbal nouns.
Just as the class of verbs divides into transitive and intransitive, the class of verbal nouns divides
into verbal noun transitive (vnt) and verbal noun intransitive (vni); some are both, for example
sicak hanta can mean either ‘begins it’ or ‘it begins’. A construction of vnt + postnominal verb can
take a direct object: Yenge lul kongpu hanta ‘studies English’. For separable verbal nouns transitive
there is sometimes an alternative way to express the object: Yenge kongpu lul hanta ‘does English
study’ (Yenge adnominal to kongpu) or ‘studies - English’ (Yenge adverbial, i.e. absolute, to the
predicate kongpu lul hanta). The separable verbal noun may take the object marker, especially if no
other object is present: kongpu lul hanta ‘does some studying, studies’. Normally, if the verbal noun
A Reference Grammar of Korean
PART I 189
is modified (by an adnoun or modifier construction) it cannot take another object and it is usually
followed by the accusative particle: elyewun kongpu lul hanta ‘does some difficult studying’,
i kongpu lul hanta ‘does this studying’.
The class of intransitive verbs has a subclass of descriptive verbs (= adjectives), characterized by
the lack of processive forms that are present for the processive intransitive (and all transitive) verbs.
Similarly, the class of intransitive verbal nouns divides into the processive (vni proper) and the
descriptive ones that we call adjectival nouns (adj-n). (For a few adjectival nouns that appear in
unexpected forms, see §7.1, p. 217.) The adjectival nouns form constructions with postnominal verbs
that are descriptive; those we call postnominal adjectives. Apparently ha- is the only common
postnominal adjective that is separable.
The lexicon should seek the following information about each verbal noun, in addition to the three
variables mentioned above:
(4) vnt # vni £ adj-n;
(5) the specific postnominal verb(s) a given verbal noun occurs with. For example, some but
not all vnt that occur with ha- also occur in a passivizing conversion with toy- (see §5.6.6).
A construction that consists of vni + postnominal verb will not take a direct object. A separable
vni is sometimes set apart from ha- by the particle ul / lul (as well as to and un / nun) or from toy- by
the particle i/ka: kyelhon (ul) hanta ‘gets married’, kekceng (i) toynta ‘gets worried’. And between
VN ul/lul and ha- you may hear one of the adverbs an ‘not’, mos ‘cannot’, or cal ‘well; lots; often’:
kongpu lul an /mos /cal hanta.
Among the verbal nouns, some are best treated as a subclass of the impressionistic (mimetic)
adverbs, §12. Many of the other verbal nouns are from the Chinese vocabulary, but there are also
verbal nouns in the core vocabulary: kancik ‘keep’, kekceng ‘bother’, kyenyang ‘aim’, kwi-ttwim
‘hint’, tacim ‘pledge’, melmi ‘feel nauseous’, pasim ‘plane’, pel.im ‘earn’, son cis ‘gesture’, ... . And
some are from the stock of modern English loans: nokhu ‘knock, hit’, tulaipu ‘drive’, ssain ‘sign,
signal’, ... .
It is difficult to decide whether to treat many of the verbal nouns borrowed from Chinese as
transitive or intransitive. Often an etymological “object” is already incorporated in the Chinese
expression (kwennong ‘encouraging agriculture’, for example, includes nong ‘agriculture’), so that it
seems pleonastic to add a separate object. Yet many of the verbal nouns listed by dictionaries as being
intransitive are used colloquially with pleonastic objects: nongsa lul kwennong hanta ‘(farming-)
promotes agriculture’.
5.6.1. Defective verbal nouns.
Defective verbal nouns are those which occur with only a few (or just one) of the expected
paradigmatic forms of the postnominal verb:
DERIVED ADVERB: MODIFIER (CF §5.3.1):
mutan hi ‘without reason/leave’ yuil han ‘unique’
chong hi ‘all, entirely, wholly’ momo han ‘prominent’
kam hi ‘with daring’ taymo han ‘main, important’
kong hi ‘alike’
kuk hi ‘extremely’
kunkun hi = kunkun (= kyewu) ‘almost’
kkun hi = ‘tenaciously, persistently’
kiyen hi ‘for sure’
congsok hi ‘without delay’
conditional:
(k)elphis hamyen ‘all too often’
tacik hamyen (= kikkes hamyen) ‘at most’
INFINITIVE + PARTICLE:
tacik hay ya ‘at most’
kupke hi ‘suddenly’ GERUND:
kosulan hi ‘intact’ (un / nun) kosa hako ‘apart (from •••)’
kaman hi ‘quietly’ kyel kho ‘definitely (not)’ < kyel(qtan) hako
kkol-kkol hi ‘sorrowfully (weeping)’ cengnyeng kho (< hako) = cengnyeng ‘definitely’
nalan hi ‘in a row’ phil.yen kho (< hako) = phil.yen ‘for sure’
190 PART I
A Reference Grammar of Korean
Those with ••• hi are obviously adjectival nouns, since hi comes only from the postnominal
adjective ha-, but it would be difficult to say for sure whether the other cases should be regarded as
processive or descriptive verbal nouns. The cases with - han have already been treated (in §5.3.1) as
quasi-adnouns, so we might do well to treat the others as types of quasi-adverb.
Certain verbal nouns occur in several but not all of the paradigmatic forms: tangmyen ‘confront’
appears only with han, hakey, and hamyen, as in tangmyen han muncey ‘the problem that confronts
us’, ilen muncey ey tangmyen hakey toymyen ‘if we come to confront such a problem’, and ilen
muncey tangmyen hamyen ‘if such a problem confronts us’. And tahayng ‘fortunate’ seems to appear
only with hakey, hi, and han; see also §5.2.9 for its occurrence with ulo and ita. Most of the verbal
nouns that occur in extended particle phrases (§6.3) are limited to infinitive (ha.ye or hay se) and
modifier (han) forms, e.g. ••• ulo in ha- ‘be due to’. But some have a few additional forms (e.g. - ey
uy ha- ‘rely upon’ occurs at least in uy hamyen, uy hanta, uy hako, and uy hamye), and some occur
only in the infinitive: - ey cuum/cey hay (se) ‘on the occasion of’.
5.6.2. Transitive verbal nouns.
Most two-syllable transitive verbal nouns that occur with the separable verb ha- are themselves
separable and occur as free nouns: kongpu ‘study’, sayngkak ‘think’, kwukyeng ‘view (for pleasure)’,
taycwung ‘estimate’, salang ‘love’, mac.i ‘meet’, ... . But hetak hanta ‘nibbles on what has been set
apart’ is an inseparable vnt. The monosyllables wen ‘desire’, mal ‘say’, and kum ‘appraise, fix the
price of’ are also separable and occur as free nouns (‘desire’, ‘worth’, and ‘price’). Most monosyllabic
vnt, however, are inseparable: cen ‘convey, report’ (related noun ‘biography’); tay ‘face, relate to’
(related noun ‘pair; versus’); cheng ‘invite’ (as noun ‘invitation’); cey ‘subtract’, hap ‘add’, kam
‘deduct’, kwu ‘get, buy’, sang ‘harm’, tang ‘undergo’, thayk ‘choose’, pong ‘seal’, yo ‘need’.
5.6.3. Intransitive verbal nouns.
Most of the two-syllable intransitive verbal nouns that occur with the separable verb ha- are
separable: kyelhon ‘get married’, sanqpo ‘take a walk’.Of the monosyllabic vni, only il ‘work’
(of the core vocabulary) appears to be separable; all the others are inseparable: kwan ‘be relevant,
related’, sok ‘belong’, ung ‘agree, consent’.
For the inseparable verbal postmodifiers chey/chek and ssa, see §5.4.3.1.
5.6.4. Adjectival nouns.
Quite a few adjectival nouns (mostly of two syllables) serve also as free nouns and are separable
from ha- by the particle to:
kanan ‘poverty / poor’ phikon ‘tiredness / tired’
kanung ‘possibility / possible’ pulhayng ‘misfortune / unfortunate’
katang ‘appropriateness / appropriate’ sicang ‘hunger / hungry’
keman ‘haughtiness / haughty’ yengwen ‘eternity / eternal’
kwung ‘destitution / destitute’ 1
1 As a free noun: kwung ey ppa cin cip ‘a house fallen into destitution’.
For the separable adjectival postmodifiers pep and tus see §5.4.3.3, and the entries in Part II.
There are also quite a few two-syllable (or longer) adjectival nouns that do not occur as free nouns
but are quasi-inseparable. In colloquial usage they are usually inseparable, but occasionally the
constructions can be split by the multivalent word tul ‘plurally’ (here treated as a particle), and in
written texts the particles to and un/nun sometimes intervene: puncwu tul hata ‘are all busy’,
puncwu to hata = puncwu haki to hata ‘is busy indeed/also’. A few of these quasi-inseparable
adjectival nouns: emaema ‘elegant’, kkaykkus ‘clean’, puncwu ‘busy’, thunthun ‘strong’.
Most adjectival nouns of one syllable are inseparable, even though a few of them occur also as
free nouns or other parts of speech: chen ‘lowly’, mos ‘be inferior’ (not to be confused with the
etymologically related adverb ‘cannot, definitely not’), sil ‘be substantial,... ’ (also bound noun, bound
adnoun), sok ‘speedy’. There are four inseparable adjectival postmodifiers. They are treated in
§5.4.3.2 (p. 161) and in the entries of Part II.
A Reference Grammar of Korean
PARTI
191
5.6.5. Bound adjectival nouns.
Some adjectival nouns are inseparable (in that they are never separated from the following ha-)
and are attached to adjective stems:
-tama, -tala (-* -tama(h)-, -tala(h)-) ‘rather (of size)’
-(k)kum ‘rather’
-(c)cek, -(c)cak, -chek ‘rather —ish’
-(c)cek-cikun, -(c)cak-cikun, -chek-cikun ‘rather •••ish’
-swuk ‘ - ish’
-swuk(-)uley ‘ - ish’
-cepcep ‘ -ish, slightly colored (tinged)’
-(k)kulum SAME
-(u)tay-tay, -(u)tey-tey, -(u)tayng-tayng, -(u)teyng-teyng SAME
-(u)chwung-chwung SAME
-(u)chik-chik SAME
-(u)thoy-thoy, -(u)thwi-thwi SAME
-upsulum ‘ -ish, slightly colored (tinged); slightly characterized by’
-us(ul)um/-s(ul)um ‘ -ish, slightly characterized by’
-(u)m(-)uley ‘ - ish’
-(u)s (see §12 and Part II)
-kkey ‘dull and ugly (colored)’
These could be called “adjectival postadjectivals”; compare the bound postverbs of §7.1.3.
Examples will be found in Part II under -usulum. Shape alternations are discussed in §12.
5.6.6. Conversion constraints on verbal nouns.
In §11.6 we describe special passive conversions for those verbal nouns of more than one syllable:
- hanta -* - toynta, -* - tang hanta, -* - pat.nunta. These conversions are limited to specific
transitive verbal nouns, which must be listed. Moreover, there are intransitive verbal nouns which also
occur with toynta, perhaps as an abbreviation of - hakey toynta. Cf hon.lan (hakey) toynta from
hdn.lan hata ‘is disarranged’, an adjectival noun. And a few verbal nouns do not occur with hanta,
but only with toynta. We can make representative lists:
(1) - hanta -*• - tang hanta
vnt kecel ‘refuse’
(2) ••• hanta -* - pat.nunta
vnt cwumok ‘watch’
hyep.pak ‘threaten’
(3) - hanta -*■ - toynta
vnt kamkum ‘imprison’
kilok ‘record’
kolye ‘consider’
kyoyang ‘educate’
vni kamtong ‘be emotionally moved’
cungtay ‘enlarge, grow’
hapkyek ‘qualify’
yongse ‘forgive’
sokay! ‘introduce (people)’
'noncung ‘prove’
sicak ‘begin’
sokay 2 ‘introduce (ideas, culture, knowledge)’
kwelki ‘be roused to action’
pensik ‘breed’
tankyel ‘unite’
192 PART I
A Reference Grammar of Korean
(4) - toynta only (and no - hanta)
vni *iik ‘prove profitable’
kachayk ‘get scolded’ mapi ‘get paralyzed’
kiceng ‘be ready-made, established’ moswun ‘be contradictory’
koco ‘reach a climax’ sangchi ‘coincide’
kyel.wen ‘become enemies’ sotuk ‘be earned’
(5) - hanta only (and no -* - toynta)
vnt kak.o ‘apprehend’
kangsup ‘assault’
kansep ‘interfere’
kyengthan ‘admire’
mulqsayk ‘search out’
vni kyekcen ‘battle’
muntap ‘quiz’
myento ‘shave’
nampok ‘dress as a man’
nolyek ‘endeavor’
pakswu ‘applaud’
Most verbal nouns of more than one syllable
— sikhinta:
vnt chwucin ‘propel’
haypang ‘liberate’
vni ciyen ‘delay’
cungka ‘increase’
haysan ‘disperse’
kamso ‘decrease’
pangkwan ‘observe as bystander’
pokcong ‘obey’
taywu ‘treat’
tokchang ‘create’
panghwang ‘wander’
pihayng ‘fly’
puncayng ‘dispute in factions’
tamhwa ‘chat’
tapcang ‘answer (in writing)’
Vehayng ‘travel’
be used in the causative conversion - hanta -»
hwaksin ‘be convinced of’
kyelhap ‘combine’
mucang ‘arm for war’
yak.hwa ‘grow weak’
For the permissible dropping of (••• hanta -*•) haci in conversions of verbal-noun sentences, see
§11.7.5, and for the stylistic dropping of (••• hanta -* ) hako, see p. 277. Dyads of like verbal nouns
(vnt + vnt, vni + vni, adj-n + adj-n) occur in construction with the postnominal verb ha-; we might
regard these as instances of an optionally omitted hako: vnj [hako] vn 2 ha-. Apparently both verbal
nouns must be of two syllables. For examples, see §10.2.
6.0. Forms: particles.
Particles are words that mark grammatical relationships, focus, emphasis, attitude, and a variety
of emotional meanings. A Korean particle follows the word or phrase which it is marking, so that the
Korean particles (like those of Japanese and many other languages) may be called postpositions, by
analogy with the prepositions of western languages, such as English. It is often difficult to translate a
given particle from one language to another, just as it is difficult to translate prepositions, which serve
a similar function in English. The translation of a Korean particle will sometimes be a preposition in
English, but it may instead involve word order or the placement of sentence stress, the choice of
definite or indefinite article, and other subtleties that are difficult to pinpoint.
6.1. Characteristics of particles.
All particles sometimes occur before pause, but it is unusual for a pause to occur before a particle,
for the particle is normally attached to the preceding word in close phonological juncture, in spite of
the fact that it is in construction with the entire preceding phrase, not just the preceding word. Most
A Reference Grammar of Korean
PART I 193
particles sometimes occur (1) after a noun; certain particles also/instead occur (2) after various
inflectional categories; and some of the particles also occur sometimes (3) after other particles. When
particles occur in sequence, as in wuli apeci EYKEY SE POTA TO ‘also/even than from our father’,
they peel off from the end, and each forms an immediate constituent with the entire preceding phrase:
wuli apeci eykey se pota + to ‘also/even’
wuli apeci eykey se + pota ‘than’
wuli apeci eykey + se ‘from/at’
wuli apeci + eykey ‘to / at a person’
(wuli ‘we/us’ + apeci ‘father’)
Most of the common particle sequences are included as entries or as subentries in Part II. An
attempt was made to elicit all conceivable sequences, including some which are merely “awkward”,
but not completely rejected. In general the focus particles (un/nun, to, iya/ya) come last in a string,
but man ‘just’ occurs in several positions, as can be seen from the examples under the entry in Part II.
The particles i/ka (nominative), ul/lul (accusative), un/nun (subdued focus), to (highlighted focus),
and iya/ya (highlighted contingency) are mutually exclusive. But the nominative particle can occur
after the other particles in certain hypostatic contexts, such as echo questions or denials (‘it isn’t [a
matter of] “ ”’), or expressions such as - i/ka muncey ’ta ‘the problem is [the matter of] “ - ”’,
into which a fragment is inserted from an assumed echo. These unusual situations are largely ignored
here, but mention is made of particle sequences of nominative following other particles (such as ey,
eykey, and ulo) under similarly limited circumstances, because those strings are more likely to be
encountered than (*)ul/lul i, (*)un/nun i, (*)to ka, or (*)iya/ya ka. Although Ton ul [ ] i muncey
’ta is barely possible for ‘The problem is [that he wants/... ] money’, there appears to be no comparable
context that would permit Ton i [ ] lul ... , except as a forced ellipsis: “Ton i” lul “ton ul” lo
kochyess.ta ‘I corrected [the phrase] “ton i” to “ton ul”’. Such special conditions would even allow
an iteration of the same particle: “Ton ul” ul “ton ulo” lo kochyess.ta ‘I corrected [the phrase] “ton
ul” to “ton ulo”’; “Wuli ka” ka muncey ’ta ‘The problem is the [phrase] “wuli ka”’.
A particle particularizes and limits the grammatical relationship of the phrase to the rest of
the sentence, places the word in perspective with respect to the rest of the sentence, or (if at the end of
the sentence) shows how the sentence is regarded with respect to the discourse or the speech situation.
6.2. Quasi-particles.
Sometimes it is difficult to decide whether what we have is noun + particle or just noun +
NOUN. There is one helpful test. The words na ‘I/me’, ce ‘I/me’ [formal], and ne ‘you’ have the
alternant shapes nay, cey, and ney before a noun but not before a particle - with the exception, for
special reasons (p. 196, note 1), of the nominative particle ka. What follows the pronominal reference
in nay sayngkak ‘my opinion’ or ‘thoughts of (= about) me’ and (ne wa) nay yeph ‘beside (you and)
me’ is a noun. For that reason we decide to treat man in na man ‘just me’ as a particle. And we can
set up a class of quasi-particle to take care of the extended use of phrases like pakk ey ‘except for’
in na pakk ey ‘except for me’ and cip pakk ey eps.ta, which has two meanings ‘there are none
outdoors’ and ‘there’s only the house’ (= cip man iss.ta). The literary synonym oy ey is to be treated
the same as pakk ey. In a sense, perhaps man could be called a quasi-particle, too, since it also serves
as an adjectival noun; but because we treat man as a particle it follows that ku is a noun, not an
adnoun, in ku man ‘just that’. Other quasi-particles are kawuntey, cwungkan, and sai ‘midst’: ne wa
na kawuntey/cwungkan/sai ‘between you and me’. In na ttaymun ‘because of me’ na is a noun and
ttaymun a postnoun (not a particle because no “other” particle can be inserted before it - see below,
§6.4), but in nay ttaymun ‘for my sake’ ttaymun is a noun.
In addition to noun quasi-particles, there are also quasi-particle verb forms which I choose not to
call particles, but treat as specialized uses of the verb itself:
(1) kath.i, pronounced /kachi/, the derived adverb of kath- ‘be like or with’. I regard na kath.i
‘like me’ as an abbreviation of na wa kath.i. The full form means either ‘like me’ (= na chelem with
a particle) or ‘with me’ (= na wa hamkkey with an adverb in reciprocal valence ‘together with’); the
194 PART I
A Reference Grammar of Korean
abbreviated form means only ‘like me’. Notice that the expression is inflected through all categories:
na kath.un ya ‘is it like me?’, na kath.ess.ta ‘it was like me’.
(2) kaciko ‘with, ... ’, the gerund of kaci- ‘hold’. See the entries kaciko and -e kaciko in Part II.
(3) all forms of the copula, especially ila/’la, ila to / ’la to, in tul / ’n tul, ina / ’na, ita (ka) / ’ta
(ka), iko / ’ko, imye / ’mye.
(4) all forms of the verb ha-, notably hako. But this word I treat as a particle proper when optionally
it substitutes for wa/kwa ‘with, and’. That is for various reasons, primarily the great frequency of
hako and its wider distribution than other forms of ha-.
(5) the abbreviation tholok (< hatolok). This functions like a postnoun of very limited distribution.
See the entry in Part II.
(6) mattana ‘according to (something said)’ < mac.ta hana ‘says it fits but’.
(7) chiko (vt gerund) ‘considering as’.
(8) malko, gerund of ma-1- ‘refrain, desist’.
Perhaps all these quasi-particles could be regarded as transitional types or “particles in the
making”. The verbal origin of several of the particles proper is still transparent: cocha, pota, puthe,
mace, mankhum .Some Korean dictionaries list as a particle the uncommon use of palo marked
“3” in the entry of Part II, but I prefer to treat it as a quasi-noun.
6.3. Extended particle phrases (phrasal postpositions).
In expository prose some of the verbs and verbal nouns are used as a way of extending and
particularizing a particle. Most of these semi-literary cliches form two kinds of phrases: adverbial,
with the infinitive or the infinitive + the particle se but occasionally with the gerund, and
ADNOMINAL, with the plain modifier. (In the table “ha.ye/hay se” means both the literary ha.ye and
the colloquial hay se are used.) The extended particle phrases function much like simple particles, so
we call them phrasal postpositions.
ADVERBIAL
ADNOMINAL
MEANING
(1)
ey kwan ha.ye / hay se
ey kwan han
‘respecting, concerning, about’
(2)
ey tay ha.ye/hay se
ey tay han
‘directed toward; treating,
concerning, regarding;
against, opposing’
(3)
ey pan ha.ye/hay se
ey pan han
‘in opposition to, contrary to’
(4)
ey pi ha.ye/hay se
ey pi han
‘compared with, relative to’
(5)
ey in ha.ye/hay se
ey in han
‘in accordance with’ Cf 26
(6)
ey uy ha.ye / hay se
ey uy han
‘depending on, based on’
(7)
ey Mm ha.ye/hay se
ey Mm han
‘facing, confronting, meeting,
in the presence of’
(8)
ey han ha.ye / hay se
ey han han
‘limited/restricted to’
(9)
ey kung ha.ye / hay se
ey kung han
‘throughout’ (= ey kelchye)
(10)
ey pichwe
—
‘in view/light of, according to’
(11)
ey ttal% (se)
—
‘consequent to’
(12)
ey iss.e (se)
—
‘in, at, for’
(13)
ey cey ha.ye
—
‘on the occasion of’
(14)
ey cuum ha.ye
—
‘on the occasion of’
(15)
ul pilos ha.ye/hay se
ul pilos han
‘beginning with, (starting) from’
(16)
ul tang ha.ye/hay se
ul tang han
‘facing, confronting’
(17)
ul hyang ha.ye / hay se
ul hyang han
‘(facing) toward’
(18)
ul thong ha.ye / hay se
ulthong han
‘through (= via or throughout)’
(19)
ul kyek ha.ye/?hay se,
ul kyek hako
ul kyek han
‘separated by; with - between;
at intervals of’
(20)
ul wi ha.ye / hay se
ul wi han
‘for the sake/purpose of, on behalf of’
A Reference Grammar of Korean
PART I 195
(21) ul ki ha.ye
(22) ul kkiko
(23) ul mak.lon hako
(24) ul pulko / pulkwu hako
(25) ulo malmiam.% (se)
(26) ulo in ha.ye/hay se
(27) ulo ha.ye/hay se
(28) ulo ha.ye-kum
(29) kwa tepul.e (se)
(30) kwa awulle (se)
ulo malmiam.un
ulo in han
‘at the time of’
‘along(side), parallel to, following’
‘to say nothing of’
‘disregarding’
‘in accord(ance) with; owing to’;
‘in consequence of’;
[with passive] = ey(key) ‘by’
‘on account of’ (Part II; CM 1:165)
‘causing/letting’ (Part II; CM 1:165)
‘together with, in common with’
‘in addition to’
An interesting fact about these expressions is that they are used after phrases that would not
otherwise take the particle. We can say, for example, that the particle ey occurs after the postmodifiers
ka and ya ‘question’, but only in such expressions as olunun ka/ya naylinun ka/ya ey ttal% (se)
‘depending on whether it is rising or falling’. In a number of other cases a particle called for by the
following expression will occur after phrases that would not otherwise attach the particle:
(31) ••• kwa/wa pantay ’ta ‘it is in opposition to’
(32) - kwa / wa pantay lo ‘in opposition to’
(33) ••• kwa/wa ilpan ita ‘it is the same as (the case of)’
(34) - kwa/wa kath.i (kath.ta) ‘(it is) the same as, like’
(35) - kwa/wa hupsa hata ‘it closely resembles (the case of)’
(36) - kwa/wa talli (taluta) ‘(it is) different from’
(37) ••• kwa / wa hamkkey ‘together with’
(38) ••• kwa / wa tongsi ey ‘at the same time (together) with’
(39) ••• ey to pulkwu hako ‘regardless of, despite’; Cf (24) above
(40) - i/ka muncey ’ta ‘it is a question; the question is (one of)’
(41) ••• i/ka ani ’ta ‘it is not (that), it is not a case of’
(42) ••• ul mak.lon hako ‘to say nothing of’; = (22) above
Compare CM 1:165, 266-7.
I have not included - ey kelchye ‘extending over (a period of time, ... )’, for it is just the
infinitive of ••• ey kelchi- ‘extend over’, which seems to occur freely in all forms. There are, of course,
quite a large number of verbs which call for the dative particle ey rather than (or as well as) the
accusative particle ul/lul. See §10.8.
6.4. Particles proper.
The list of particles that follows is partly ordered by semantic groups, but I have omitted tag
translations since they would be misleading. The full range of meaning (often extensive) and use (often
overlapping) will be found in the entry for each particle in Part II. I have indicated a few strings that
are often treated as single particles by Korean grammarians, such as ey se. The sequences that have
actually been found or elicited, together with further distributional limitations, are given in §6.5-6 and
individually in Part II.
1. i/kal
8.
tele
2. ul / lul 2
9.
kkaci
3. ulo / lo
10.
se
ulo se / lo se
ey se
ulo sse / lo sse
eykey se, hanthey se; kkey se
4. ey
11.
puthe
ey ’ta (ka)
12.
iya / ya
5. eykey; [honorific] kkey
13.
man (also adjectival noun)
6. hanthey
14.
mankhum, manchi
7. poko
15.
khenyeng
196 PART I
A Reference Gra mm ar of Korean
16. mata
17. ssik
18. ccum
19. un / nun 3
20. to
21. cocha
22. mace (variant maca)
23. pota (also [written-style] adverb)
24. chelem
25. kwa / wa
26. hako
27. sekken
28. ilang / lang (= sekken; = kwa / wa)
29. ullang / Hang (= un / nun)
30. ko (after verb forms in indirect quotations)
31. iyo / yo
32. kwulye (after verb forms)
33. kkwuna (as particle, only after -ca)
34. uy
35. ci (= uy in Chinese cliches)
36. a / ya
37. una / na (after interjections, -key, -ulyem)
38. la (only after -^a)
39. tul (also postnoun, adverb)
40. son (only after -ta; also postmod) [obsolete]
41. ppun (only after -ta; also postnoun, postmod)
42. nayci (NUMBER! ~ Number 2 )
43. (-ulq ci) enceng
44. (-ki nun) saylo (ey) / say lye
1 In the standard language i follows a consonant and the suppletive ka follows a vowel. But in some
northern dialects i or its reduction y occurs after vowels, too, and that is the regular pattern found in
the early Hankul texts. There are cases, especially in the north, of pleonastic - i ka (but no *••• ka ka
or *••• ka i) which suggest that the ka may have been added for emphasis; the standard nay ka ‘I’ for
na ka (a common dialect version) probably represents that formation, as do ney ka ‘you’ (dialect ne
ka), cey ka ‘I [formal]’ (dialect ce ka) and dialect nwi ka for standard nwu ka ‘who’. The 15th-
century forms of these words were nay (1447 Sek 6:14b, 6:19b, 6:24:29b) < na i, "ney < ne i
(1447 Sek 6:1a), "cey (1447 Sek 9:14a, 9:21a - correcting Martin 1988, which copied a misprint in
NKW) < ce i, and "nwuy (1449 Kok 36) < nwu i.
2 In Cincwu (South Kyengsang) lo is used for lul but ul is never said as ulo (Mkk 1960:3:31). In
parts of the north people drop the final liquid and say lu for lul and u for ul. Some of the northern
dialects use (u)lwu or (u)lu for (u)lo, so that for certain speakers the form used for the standard ul / lul
may converge with that used for ulo/lo after a vowel or the liquid. The shape lul readily abbreviates
to just 1’ and both versions coexisted in the first Hankul texts and apparently also in the language of the
12th century and earlier, if our interpretations of the hyangka orthography are correct. See note 3.
3 The shape nun readily abbreviates to just n’ and both versions coexisted in the first Hankul texts
and earlier materials, as did 1’ and lul. For both these particles it is usually assumed that the short form
represents the original morpheme and the full form iterates (reduplicates) that. Cf Kim Wancin 1975.
The last three cases (42, 43, 44) are somewhat anomalous. Apparently they are always in close
juncture with the preceding phrase, so they are treated by most Korean dictionaries as particles rather
than, say, as bound nouns or adverbs. Compare the note on kot in Part II.
Nouns of relative location, such as wi ‘atop, above’, alay ‘below’, and yeph ‘beside’, are not
treated as particles. In chayk-sang wi ey ‘on top of the desk’ only the ey is a particle, and the
preceding expression is noun + noun for two reasons:
(1) we can say just wi ey ‘on top’ without a preceding noun (but of course ‘on top’ implies ‘on top
of something’), and wi occurs in other positions as a free noun, though not modified by an adnoun or
verb modifier (so that in ku wi ‘on top of that’ ku is a noun, as it is in ku man ‘just that’);
(2) ‘above me’ is nay wi ey not *na wi ey. The same is true of yeph (ey) ‘beside’: nay yeph (ey)
‘beside me’ shows that yeph is a noun even though tangsinq yeph ‘beside you’ and wuli yeph ‘beside
us’ might leave one wondering whether it is not a postnoun or particle. There are dialects (such as that
of Taykwu) in which na yeph, na wi, etc., are used instead of the forms with nay, but speakers of
those dialects say na chayk for ‘my book’. For such dialects, the fact that wi and yeph can begin a
sentence is sufficient criterion to establish that they are not postnouns or particles. Nouns of relative
location can be thought of as distilled from a kind of semantic predication that locates one noun with
A Reference Grammar of Korean
PART I 197
respect to another. Other such nouns: an ‘inside’ (something rather empty), sok ‘(deep) inside’
(something rather full), kawuntey ‘inside; between, among’, sai ‘between, among; midst, during’,
cwung ‘midst, among’, cwuwi ‘around (the periphery of)’, kyeth ‘beside, in the vicinity of’, mith
‘(at) the base/bottom of; beneath’, aph ‘in front of; (= cen) before, ahead of’, twi ‘behind; (= hwu)
after, later than’, ... . By their meaning (‘between, among’) some of these words locate a noun with
respect to two or more other nouns.
Moreover each of the particles listed as occurring with nouns is sometimes preceded by at least
one other member of the same list. That is the ultimate distributional fact that determines the list.
6.5. Particle sequences.
Particles occur in sequences of two (eykey se), three (eykey se nun), and even four (ey se puthe
uy) or, rarely, more (ccum ey se pota to, ... ). The longer sequences usually end with one of the
particles un/nun, to, or uy. In my analysis the constituency cut is always between the particle on the
right and the remainder of the phrase, so that each particle “peels off” in turn from the right. Yet even
though this analysis does not treat the particles in sequence as in construction with each other, it is
interesting to see what sequences can be found. I have looked for all possible shorter combinations and
tried to elicit those which I could not find in texts. Some sequences alleged to exist (e.g. by CM) are
rejected because I have been unable to elicit satisfactory examples. Among these are:
?* chelem iya mal lo; (ey) se lul; (ey) se lo, ey wa; ina to,
pota ’na, pota uy, pota chelem, pota kkaci, pota ya mal lo,
man iya mal lo, tele ’na-ma to, tele uy, ulo lul, ulo wa.
The sequences that have been found are supplied with substantiating examples in the entries of Part II.
Here part of that information is recapitulated in a different form.
Ignoring for the moment certain synonyms, such as the colloquial synonym hanthey for eykey,
and less interesting forms, such as poko and tele for only some of the uses of eykey, we will examine
the sequences of particles that have been found to occur after nouns and include the copula forms ila
to, ina(-ma), in tul, itun ci, and (i)ta (ka) as if they were particles. Some of the strings are rare, and
some of the examples are extremely colloquial. Certain sequences could be elicited only in sentences
that speakers considered “awkward”. Undoubtedly there are sequences that have been missed; in
particular, I would expect to find more with final un/nun and to, since I did not try to elicit all
possible longer sequences with those two particles. Some of the sequences with i/ka as the last
member are obtainable only with the copula negative construction - i/ka ani ’ta (SEE remarks at end
of §11.7.2) and these are marked “n”. In the lists I have used the designations eykey, (k)wa, and
sekken to include sequences that were actually found with the more common Seoul colloquial forms
hanthey, hako, and ilang/lang (respectively). The more exact information given in Part II suggests
that a few of the colloquial versions, because of the relative infrequency of the sequence, are rejected
in favor of the less colloquial synonyms; see, for example, the notes on man kwa ( -*■ man hako),
kwa kkaci (-*■ hako kkaci). Although hako occurs more freely than kwa/wa, certain extended
particle phrases (§6.3), because of their stiff and literary flavor, take only kwa/wa: you do not hear
hako taking the place of the less colloquial particle in - kwa/wa tongsi ey ‘at the same time
(together) with ••• ’.
In the two lists that follow, the sequences are given alphabetically: in the first list by the prior
member, in the second by the latter member. Each particle is listed by its postconsonantal shape even
though the other shape might be appropriate to the particular sequence, so that eykey + (k)wa =
eykey wa, eykey + i/ka = eykey ka. Space within two-word units is here shown as
It will be seen that virtually no other particle ever follows these markers:
i /ka (But there are examples of i / ka tul in Part II. And see -ta ka ka in §6.6; this may be an
argument against that analysis.)
ul / lul (But there are examples of ul tul in Part II.)
198 PART I
A Reference Grammar of Korean
uy
to (But there are examples of to tul in Part II.)
iya (mal lo)
ilato
ina (But there are examples of ina lul and ina tul in Part II.)
ina-ma
ita_(ka) (But there is an example of ’ta ka tul in Part II.) Contrast -ta (ka), which can be
followed by (n)un, to, tul, (i)ya, and even ka.
itun ci
The particle un / nun is followed only by khenyeng or by tul, and khenyeng is followed only by
un/nun or by tul. That means it is possible to get - un/nun khenyeng un, as in the following
(perhaps somewhat unlikely) sentence: Cel.yak ulo ey ya mal lo nun khenyeng (un) hana to eps.ta
‘Far from being a matter of economizing on them, I just haven’t got a single one’. The particle ssik is
preceded only by ccum; the particle mankhum is preceded only by chelem, eykey, and kkaci. In older
usage, mata is preceded only by ccum but modern usage prefers the order eykey mata, (ey) se mata,
and eykey se mata (kkey se mata) to the older usage with mata first; and rather than mata ey the
modern usage prefers simply mata. That leaves ssik eykey and perhaps ccum eykey as the only
sequences with a particle preceding eykey.
Every particle that occurs after a noun is sometimes preceded by at least one other particle. That
criterion alone is sufficient to distinguish a particle from a postnoun.
The list of particles excludes the sequence kkey se (honorific oblique subject); it has the same
following partners as se alone, and additionally also i/ka, mata, and sekken. In Seoul sekken is
usually replaced by the synonym ilang; I have assumed that the distribution is the same in the same
meaning. (The particle ilang is also used as a synonym of hako = kwa/wa ‘with’.) Like eykey (se),
the sequence kkey se is preceded only by the particles ssik and mata, and modern usage refers kkey se
mata. Not included are strings with ppun (man) ‘only’ such as ulo se ppun and ey se ppun man,
which will be found in Part II. Notice also (in Part II) the unusual sequences ulo sse (nun) and iya mal
lo, as well as the written-style strings ey iss.e se (’na, nun, uy).
There are pairs in contrasting or competing order:
cocha mace
(mace cocha)
ccum chelem
chelem ccum
ccum eykey
eykey ccum
ccum ssik
ssik ccum
ey cocha
cocha ey [rare]
ey kkaci
kkaci ey
ey man
man ey
eykey ccum
ccum eykey
se (k)wa
(k)wa se [rare]
ulo kkaci
kkaci lo
ulo man
man ulo
kwa man
man kwa ( -*•
man hako)
We can add the cases with mata mentioned earlier:
eykey mata
ey se mata
eykey se mata
kkey se mata
There are longer sequences such as:
kkaci ey se mace
(mata eykey)
(mata ey se)
(mata eykey se)
(mata kkey se)
ey se kkaci mace.
A Reference Grammar of Korean
PARTI
199
6.5.1. List of particle sequences arranged by prior member.
ccum -
chelem -
chelem
chelem iya
cocha
cocha to
ey
ey (i)na
eykey
i (/ ka)
ilato
ina
ina-ma
intul
iya
ey se
ey se pota
kkaci
?kkaci (i /) ka
kkaci (i)ya
kkaci man
kkaci (n)un
kkaci to
mace
mace to
man
man ila to
man un
mankhum
mata
mankhum ina
pota
pota (n)un
pota to
puthe
puthe (i I) ka
puthe (n)un
puthe se
se
se pota
se pota (n)un
se puthe
se puthe (i /) ka
se puthe (n)un
ssik
ssik ila to
ssik ina(-ma)
ssik man
ssik man ila to
to
ul
ulo
un
uy
ssik ulo
ssik un
ssik ulo to
ccum
ccum ila to
ccum ina(-ma)
ccum in tul
ccum iya
ccum man
ccum man un
ccum kkaci
ccum kkaci ka
ccum kkaci (i)la
ccum kkaci (i)n
ey se pota to
200 PART I
A Reference Grammar of Korean
ccum kkaci (i)ya
ccum kkaci man ccum kkaci man un
cocha ccum kkaci to
ilato
ina(-ma)
in tul
iya
khenyeng
kkaci kkaci (i)ya
kkaci (n)un
mace mace to
man man ila to
man to
man un
mankhum mankhum ila to
mankhum ina(-ma)
mankhum in tul
mankhum iya
mankhum man
mankhum to
mankhum un
pota pota (n)un
puthe
to
cocha - ey [rare]
(i /) ka
ilato
(i)na(-ma)
(i)ya
mace
?puthe (i /) ka
to
(Dul
(и) lo
(n)un
?uy
ey - chelem
cocha
(i/)ka
(i)la_to
(i)na(-)ma
(i)n_tul
(i)ta(_ka)
(')ya
khenyeng
kkaci
(к) wa
A Reference Grammar of Korean
PART I 201
eykey -
mace
man
pota
puthe
se
se chelem
se cocha
se (i /) ka N
se (i)na(-ma)
se (i)n_tul
se (i)ya
se khenyeng
se kkaci
se (k)wa
se kkaci mace
se mace
se mace puthe
se man
se mankhum
se mata
se pota
se man ila to
se puthe
se sekken
se to
se tul
se puthe (i / )ka
se puthe to
se puthe (n)un
se puthe uy
se (n)un
se uy
se (n)un khenyeng
sekken
to
(Dul
(u)lo
(u)lo uy
(n)un
(n)un khenyeng
uy
ccum
ccum iya
chelem
cocha
(i /) ka N
(i)la_to
(i)na(-ma)
(i)n_tul
(i)ta(_ka)
(i)ya (i)ya tul
khenyeng
kkaci
(k)wa
mace
man man i( / ka)
man ila to
?man to
?man un
mankhum mankhum man mankhum man un
202
PARTI
A Reference Grammar of Korean
kkaci -
mata
pota
puthe
pota to
se
se chelem
se cocha
se (i)ka
se (i)la_to
se (i)na(-ma)
se (i)n tul
(i)ta(_ka)
se (i)ya
se (k)wa
se mace
se chelem man
se man
se mata
se man to
se man un
se pota
se pota to
se puthe
se to
se tul
se puthe (i /) ka
sekken
to
tul
(l)ul
se (n)un
(u)lo
(n)un
uy
(u)lo puthe
(u)lo uy
ccum
ccum un
cocha
cocha to
ey
ey se
(i/)ka
(i)la_to
(i)na(-ma)
?(i)n_tul
(i)tun_ci
ey (n)un
ey se mace
(i)ya
khenyeng
mace
(i)ya mal lo
man
se
to
tul
(l)ul
(u)lo
(n)un
uy
man ila to
A Reference Grammar of Korean
PARTI
203
kwa - ccum
chelem
cocha
ey
(i /) ka
(i)la_to
(i)na
(i)ya
kkaci
mace
man
pota
puthe
se [rare]
to
tul
(l)ul
(u)lo [rare]
(n)un
uy
mace - (cocha)
(i /) ka
(i)la_to
(i)na-ma
(i)n_tul
?(i)ya
man
puthe puthe (i /) ka
to
(Dul
(n)un
uy
man - ey
i (/ ka)
ila to
ina(-ma)
in tul
iya
?kwa-»hako
to
ul
ulo ulo nun
ulo uy
un
uy
chelem man
kkaci (l)ul
mace to
?(man ulo)
man un
(u)lo nun
man ulo (n)un
204 PART I
A Reference Grammar of Korean
mankhum -
mata -
pota -
puthe -
(cocha)
i (/ ka)
ilato
ina(-ma)
?in_tul
iya
man man ila to
ssik
to
ul
un
(ey) (ey se)
(eykey) (eykey se)
(*)(i /) ka
(i)na
(i)ya
khenyeng
kkaci
(k) wa
man
pota
puthe
se
(?)to
(l) ul
(u)lo
(n)un (n)un khenyeng
uy
(i)la_to
?(i)na-ma
(i)ya (i)ya mal lo
khenyeng
man
to
(n)un
uy
cocha
(i /) ka
(i)la_to
(i)na(-ma)
(i)ya
khenyeng
man
pota
se
to
tul
A Reference Grammar of Korean
PARTI
205
se -
Note:
sekken -
ssik -
(Dul
(n)un
uy
chelem
cocha
(i /) ka N
(i)la_to
(i)na(-ma)
(i)n_tul
(i)tun_ci
(i)ya
khenyeng
kkaci
(k)wa
mace
man
mankhum
pota
puthe puthe (n)un
sekken
to
tul
(n)un (n)un khenyeng
uy
Also found are kkey se ka and kkey se sekken.
ccum ccum iya
cocha
iya
khenyeng
kkaci
mace
man
pota pota to
puthe
tul
ul
un un khenyeng
uy [rare]
ccum
chelem
cocha
eykey
i (/ ka)
ilato
ina(-ma)
in tul
iya
khenyeng
kkaci kkaci (i)la_to
kkaci (n)un
206 PART I
A Reference Grammar of Korean
ulo -
6.5.2. List of
- ccum
- chelem
- cocha
kwa
mace
pota
puthe
to
ul
ulo
un un khenyeng
uy
(i /) ka N
(i)la to
(i)na(-ma)
(i)ya (i)ya mallo
kkaci
mace
man
pota
puthe puthe (i /) ka
se se mankhum
se (i)ya
se (i /) ka
se pota se pota to
se to
se (n)un
se uy
sekken
to
(n)un
uy
particle sequences arranged by latter member.
chelem
eykey
kkaci
kwa
sekken
ssik
ccum
ey
eykey
kwa
mankhum
se ey se
eykey se
ssik
ccum
chelem
A Reference Grammar of Korean
PARTI
207
ey
eykey
kkaci
kwa
(mace)
puthe
se ey se
eykey se
sekken
ssik
ulo
- ey
ccum
cocha [rare]
kkaci
kwa
man
- eykey mata
(ccum)
(mata)
ssik
- i / ka
ccum
cocha
ey
eykey N
kkaci
?ccum kkaci
kwa
mace
man
eykey man
mankhum
mata
puthe
ccum puthe
(cocha puthe)
eykey puthe
mace puthe
se puthe
ulo puthe
se N
ey se N
ssik
ulo N
Note: Also found is kkey se ka.
- ila to ccum chelem ccum
chelem
cocha
ey
eykey
kkaci ?ccum kkaci
ey se puthe
eykey se puthe
chelem ccum kkaci
208 PART I
A Reference Grammar of Korean
kwa
mace
man ccum man
kkaci man
mankhum man
ssik man
mankhum
pota
puthe
se ey se
eykey se
ssik ccum ssik
ulo
- (i)na ccum
chelem
cocha
?ey ccum ey
eykey
kkaci
kwa
man
mankhum ccum mankhum
mata
puthe
se
ssik
ulo
- (i)na-ma ccum
chelem
ey
eykey
kkaci
mace
man eykey man
pota
puthe
se ey se
eykey se
ssik
ulo
- (i)n tul ccum chelem ccum
chelem
ey
eykey
?kkaci ccum kkaci
mace
man
?mankhum chelem mankhum
A Reference Grammar of Korean
PARTI
209
- (i)ta(_ka)
- (i)tun_ci
- (i)ya
- khenyeng
puthe
se ey se
eykey se
ssik
ey
eykey
ulo
kkaci
se
ccum chelem ccum
eykey ccum
chelem
cocha
ey
eykey ccum eykey
kkaci ccum kkaci
kwa
mace
man
mankhum
mata
pota
puthe
se ey se
eykey se
sekken
ssik
ulo
chelem
ey
eykey
kkaci
mata
pota
puthe
se
ey se
sekken
ssik
un
ey (n)un
mata (n)un
se (n)un
sekken un
ssik un
ey se (n)un
210 PART I
A Reference Grammar of Korean
- kkaci
- (k)wa
- mace
- mankhum
- mata
- pota
ccum
chelem
ey
eykey
kwa
mata
se ey se
sekken
ssik
ey
eykey
man
mata
se ey se
eykey se
ssik
ccum
chelem
cocha
ey
eykey
kkaci
kwa
se
ey se
eykey se
sekken
ssik
ulo
ccum
chelem
se ey se
ulo se
ccum
ey
eykey
ey se
ulo se
ccum
ey
eykey
kwa
mata
puthe
se ey se
sekken
kkaci ey se
A Reference Grammar of Korean
PARTI
211
ssik
ulo
- puthe
ccum
ey
eykey
kwa
mace
mata
se
sekken
ssik
ulo
ey se
- se
ccum
ey
ccum ey
mata ey
eykey
kwa [rare]
mata
puthe
ulo
- sekken
ey
eykey
se
ey se
ulo
- ssik
ccum
- to
ccum
chelem
cocha
ccum cocha
kkaci cocha
mankhum cocha
ey
eykey
kkaci
ccum kkaci
kwa
mace
ccum mace
kwa mace
man
eykey man
mankhum
mata
pota
ccum pota
se pota
sekken pota
puthe
se
ey se
eykey se
ulo se
ey se mace
chelem ccum kkaci
eykey se man
ey se pota
ccum ey se pota
212 PART I
A Reference Grammar of Korean
- (Dul
- (u)lo
- (n)un
ssik
ulo ssik ulo
ccum
cocha
ey
eykey
kkaci kwa kkaci
kwan
mace
man
mankhum
mata
puthe
sekken
ssik
ccum
cocha
ey
eykey
kkaci
kwa [rare]
man
mata
ssik
ccum
chelem
cocha
ey
eykey
kkaci
kwa
mace
man
mankhum
mata
pota
se pota
puthe
ccum puthe
se puthe
se
ey se
eykey se
ulo se
sekken
ssik
ccum ssik
?(kwa man)
chelem ccum
kkaci ccum
kkaci ey
ccum kkaci
ssik kkaci
ccum man
kwa man
eykey se man
ey se puthe
A Reference Grammar of Korean
PARTI
213
ulo
- uy ccum
cocha
ey
eykey
kkaci
kwa
mace
man
mata
pota
puthe
se
sekken [rare]
ssik
ulo ey (u)lo
eykey (u)lo
man ulo
6.6. Sequences of ending + particle.
Some of the particles listed in §6.4 occur in quite limited environments, and a few are found only
after verb forms. For example, una/na is found only after -ulyem, -key, and -sey; ppun and son (as
particles) only after -ta; kkwuna only after -ca; kwulye (as particle) only after statement forms
-sup.nita and -ney (and ilq sey = iney), suggestion forms -upsita and -sey, command forms -usio and
-key; la only after the infinitive The particle that marks indirect quotation ko ‘(saying/thinking
that ... ’) is used after plain quotation forms -(nun)ta ‘does/is’, -tula ‘was (doing it), I recall’, -ca
‘let’s’, -la ‘do!’; after the intentive -ulye ‘intending/wanting to do’; after the adjunctives -nula and
-ulla; and after the postmodifier ya ‘question’ that appears in -un ya, -nun ya, -tun ya.Only a few
verb forms are never followed by a particle, notably the sequential -uni, for the strings /-unipota/ and
/-unimankhum/ represent -un i pota/mankhum with the particle attached to a modifier + the
postmodifier i ‘fact (that -■ )’. Certain endings are followed by only a few particles or quasi-particles:
-ule/-le is followed by (un/)nun, to, tul, and (ul/)lul;
-una/-na is followed by tul and awkwardly by (i)la to and (i)tun ci (see also ina tul in Part II);
-una-ma is followed by yo and somewhat awkwardly by to;
-(nun)ta is followed by ko, tul, and man;
-ney is followed by tul, man, and kwulye;
-so/-o is followed by tul and man;
-kwun is followed by a;
-keni is followed by (kwa/)wa.
The particle tul ‘(acting) severally’ can be inserted rather freely in verb phrases, but not even it
can be inserted into the fixed sequences -e la, -ta ppun, -ta kwulye, and -ca kkwuna. Where tul goes
is after the other particles in those cases, as well as in -e yo (tul) ‘do/are severally’ [polite] and -ta ko
(tul) hanta ‘says that they do/are’. Although -(sup)nita tul occurs, it cannot be quoted as such, even
directly; there is no *-ta tul hako (or ila ko) hanta. And notice that it is Noun tul iyo rather than
*Noun iyo tul in fragment sentences such as Nwukwu tul iyo ‘Who all?’ [polite].
Verb forms have not been found before the particles eykey ‘to (a person)’, ssik ‘each’, or sekken
‘and the like’, nor before the quasi-particle ina-ma ‘at least, even’. Of these, sekken is the only one
that is likely to turn up, if we keep looking. I was able to elicit the particle mata ‘every’ only after
kwa (u)lo
man ulo kwa man ulo
se puthe ey se puthe
ey se
214 PART I
A Reference Grammar of Korean
-umye, and in an awkward example at that. The particle uy ‘of’ was elicitable only after the summative
-ki; the particle (k)wa ‘with’ only after -ki or in extended particle phrases (§6.3) after the substantive
ending -um. The poorer distribution of -um is unsurprising, for it is less colloquial than -ki.
The table on p. 215 shows the sequences found or elicited for most of the particles and quasi¬
particles after common verb endings. Sequences for which I have only awkward examples are indicated
by “x” instead of “ + ”. Sequences with the particle i/ka that seem to be available only when followed
by the negative copula are noted by “n”. The table is so arranged that the particles form a vertical
axis, and are arrayed alphabetically by the postconsonantal shape as in earlier lists. The endings form
the horizontal axis, and at the appropriate intersection each sequence discovered or elicited is indicated
by a “ + ”, those unfound by a Notice that the list is largely limited to sequences of just ONE
particle after an ending; there are longer sequences and they are not limited to those here shown in
more detail for se only. We can find not only -ki ey but also -ki ey to, -ki ey nun, -ki ey ya, -ki ey
tul, -ki ey pota, -ki ey sekken; not only -ci yo but -ci yo man and -ci yo man un. And many cases of
ending + man can be followed by un or to.
iS.7. Some consequences of particle distribution.
When we look at the particle sequences found after nouns and after verb forms, certain things
come to our attention. The particle se is frequently preceded by the particle ey after nouns; often the
same expression may be said with or without the intervening ey, the meaning unchanged. But in some
cases, including the summative -ki and the substantive -um, the expression is awkward (or unheard of)
without the intervening ey. In other cases, such as the verb forms -e se, -ko se, and -umyen se, it is
Ihe intervening ey that is unheard of. These facts, I think, are what leads the Korean grammarians (in
disregard of eykey se and hanthey se) to set up a single particle eyse and treat se (after a noun) as an
abbreviation of that particle. The se which occurs after verb endings is treated either as a separate
particle or as part of an unanalyzed ending that is distinct from the form without the particle. (A case
could be made that -e is an abbreviation of -e se, but no one has claimed that, so far as I know.) We
can also see why the Korean grammarians would hesitate to accept la in -e la as a particle, since
nothing can intervene between the infinitive and the particle, and la occurs as a particle only after the
infinitive, which actually turns up in a variant “pre-la” shape after certain stems: one la for wa la,
kake la for ka la, iss.ke la for iss.e la, ... (§8.3.6). In fact, -e la developed from MK - '(k)e-'la with
the imperative ending attached to the bound stem that marked the effective aspect, and one la < 'wo¬
rn- la contains a suppletive form of that aspect marker.
The grammarians’ preferred treatment of some of the uses of -ki ey, -ki lo, -um ey (= -umay),
and -um ulo as single unanalyzable endings, distinct from ending + particle, is supported by certain of
the particle sequences. And their preference for treating the transferentive -ta as an abbreviation of an
uinanalyzed ending -taka, rather than analyzing that as the assertive -ta -I- particle ka (as it is here
viewed), is perhaps supported by the recurrence of the particle ka in the sequence -ta ka ka, even
though that is a by-product of the negative copula. (There are historical arguments against the analysis
of -ta ka, too, but the matter remains open. See §9.8 and Part II; for MK accent evidence, see p. 71.)
The treatment of certain forms of the copula as particles is also supported by particle sequences.
The “quasi-particles” included in the sequence tables occur freely in positions where other forms of the
copula would be unusual or impossible. (We are, of course, here excluding from our discussions of
distribution certain kinds of peripheral utterances, such as those that occur when talking about
words.)
It will be noticed that certain “particles” after verb endings (kwulye, kkwuna, una/na, la, son,
ppun) differ from particles after nouns in that no other particle is insertible, and that mutual
insertibility was our ultimate criterion for being sure that an element after a noun really is a particle. In
a sense, what we do is to establish a class of noun particles on that basis, then we notice that some of
these particles (se, iya/ya, iyo/yo, ... ) also occur after verb endings, with greater restrictions, and
finally we discover that a few other elements, still more restricted, seem to belong in the same class,
because they occur after verb endings that are otherwise free.
A Reference Grammar of Korean
PART I 215
Table of verb endings + particles and quasi-particles
ccum
-ca
-ci
-e
+
-key
+
-ki -
+
ko
X
-ta ka
-tolok
+
-tula
-ulye
-umye
-umyen
-um
chelem
-
-
-
-
+
-
+
-
-
-
-
—
-
cocha
X
-
+
+
+
X
+
+
-
+
+
-
+
ey
—
—
—
—
+
—
-
—
—
—
—
-
+
i/ka
N
+
-
+
+
+
N
N
-
-
N
N
+
(i)la_to
-
-
+
+
+
+
+
-
-
+
+
-
+
(i)na
-
+
+
+
+
+
+
+
-
+
+
+
-
(i)n_tul
-
?
+
+
+
+
+
+
-
+
+
+
+
(i)ta_ka
-
-
+
-
-
-
-
-
-
-
-
-
-
(i)tun_ci
-
-
+
-
-
+
+
+
-
+
+
-
-
(i)ya
+
+
+
+
+
+
+
+
-
+
+
+
-
(i)yo
-
+
+
+
+
+
+
-
-
-
-
-
+
khenyeng
+
-
-
+
+
+
+
+
-
+
+
+
+
kkaci
+
+
+
+
+
+
+
+
-
+
+
-
-
kwa / wa
-
-
-
-
+
-
-
-
-
-
-
-
-
mace
-
+
+
+
+
+
+
-
+
+
-
-
man
+
+
+
+
+
+
+
+
+
+
+
-
-
mankhum
-
-
-
-
+
-
-
+
-
-
-
-
-
pota
-
-
-
-
+
+
-
+
-
-
+
-
+
puthe
+
+
+
+
+
+
-
+
-
+
+
-
+
se
-
-
+
-
+
+
+
-
-
+
-
+
_l
se ka
-
-
N
-
-
+
-
-
-
-
+
N
-
se ’lato
-
-
+
-
-
+
-
-
-
+
+
-
-
se ’na
-
-
+
-
-2
+
-
-
-
-
-
+
-
se ’ntul
-
-
+
-
-2
?
-
-
-
-
-
N
-
se ’tun ci
-
-
+
-
-2
+
-
-
-
-
-
-
-
se ya
-
-
+
-
-2
+
-
-
-
-
-
+
-
se man
-
-
+
-
-2
+
-
-
-
+
-
+
-
se puthe
-
-
+
-
_2
+
-
-
-
-
-
+
-
se to
-
-
+
-
-2
+
-
-
-
-
-
+
-
se nun
-
-
+
-
-2
+
+
-
-
-
-
+ 1 2 3
-
to
-
+
+
+
+
+
+
+
+
+
+
-
+
tul
+
+
+
+
+
+
+
+
+
+
+
+
-
ul/lul
—
+
+
+
+
+
-
-
-
+
-
-
+
ulo / lo
-
-
-
-
+
-
-
-
-
-
-
-
+
un/nun
—
+
+
+
+
+
+
+
—
+
+
+
+
1 But -urn ey se (•••) occurs.
2 Less awkward with ey inserted, but unlikely even so.
3 Also -umyen se kkaci, -umyen se tul.
216 PART I
A Reference Grammar of Korean
7.0. Forms: verbs.
In the following sections we examine various kinds of verbs (§7.1), bound verbs (§7.2), verbs
with defective paradigms (§7.3), derived stems that are causative or passive (§7.4), auxiliary verbs
(§7.5), and postnominal verbs (§7.6).
7.1. Kinds of verbs.
Each verb may be classed as transitive (vt) or intransitive (vi), though a few stems serve both
functions: kuchinta (vi/vt) means either ‘ends it’ or ‘it ends’. A verb that is sometimes preceded by a
direct object is transitive. All transitive verbs are also processive so that the opposing category of
descriptive is relevant only for intransitive verbs.
A semantic direct object may be marked by the accusative particle ul/lul, but that particle is also
used to mark other roles. If we use the cooccurrence of an accusative-marked phrase to categorize
“transitive” verbs, we will have to recognize what I have called pseudo-intransitive verbs (pseudo-
vi), which function like those intransitive verbs that are seldom, if ever, preceded by an accusative-
marked phrase. The accusative phrase could still be treated as an object, but the object is limited to
certain kinds:
(1) a noun showing the destination: hak.kyo lul kanta = hak.kyo ey kanta ‘goes to school,
attends school’.
(2) a noun showing the purpose: kwukyeng ul kanta = kwukyeng ul hale kanta ‘goes to see it’;
(3) as an object complement, the substantive form of the same stem (or other semantically and/or
etymologically cognate objects): cam ul canta = canta ‘sleeps (one’s sleep)’, chwum ul chwunta =
chwunta ‘dances (a dance)’, kel.um ul ket.nunta = ket.nunta ‘walks (one’s steps)’, kkwum ul
kkwunta ‘dreams (a dream)’, wus.um ul wus.nunta ‘laughs (a laugh)’ (Cf salam uy musik ul
wus.nunta ‘laughs at a person’s ignorance’, with a transitive use of the verb); kichom kis.kwo (1608
Twu-cip 1:10b; < kich(i)-) ‘coughing (a cough)’. But notice that kulim ul kulinta ‘draws (a drawing),
paints (a painting)’ and cim ul cinta ‘bears a burden’ are transitive verbs + tangible objects. There are
also a few cognate subjects: noph.i ka noph.ta ‘is high (in height)’, kiph.i ka kiph.ta ‘is deep (in
depth)’, nelp.i ka nelp.ta ‘is wide (in width)’, khi ka khuta ‘is big (in bigness) = is tall (in stature)’,
kil.i ka kllta ‘is long (in length)’.
(4) an expression of duration: sahul ul onta ‘comes for three days’.
(5) an expression of order or number of times: ches-ccay lul kanta ‘goes first’, sey pen ul kanta
‘goes three times’.
(6) a place traversed or a path traveled: kang ul kenne kanta ‘goes across the river’, kil ul kel.e
kanta ‘walks (along) the road’, unhayng ul cina kanta ‘passes the bank’.
(7) a quantified distance traversed: chen-li lul ttwinta ‘leaps a thousand leagues’.
(8) an affected part of the body: tali lul centa ‘limps (in a leg)’.
(9) substituting for some other particle in an unusual paraphrase: swul ul chwi hanta = swul ey
chwi hanta ‘gets drunk on liquor’. (But many, I am told, reject ul and prefer swul i chwi hanta).
Quite a few intransitive verbs take accusative-marked objects in specialized figurative senses, such as
su- ‘stand’ in aph-cang ul su- ‘stand in the van (in the lead)’ and Keki 1’ susye ya toykeyss.nun ya
‘Must you stand there?’ Usually we can say that ul/lul in these expressions is simply substituting for
some other particle, typically ey, or for a larger construction.
A verb is intransitive if it is never preceded by a (semantic) direct object: nuc- ‘be late’, cwuk-
‘die’, anc- ‘sit down’, iss- ‘stay’. Each of the intransitive verbs falls into one of the following classes.
A processive verb lacks the category of plain indicative assertive -ta, replacing it by the processive
assertive -nunta/-nta, except in literary Korean and in set literary phrases used in the colloquial. All
transitive verbs are processive, but some of the intransitives are descriptive verbs - here called
adjectives (adj). The reference is not so broad as suggested by the corresponding English category,
which includes not only predicated adjectives but also many attributive terms that are treated as nouns
or adnouns in Korean.
A Reference Grammar of Korean
PARTI 217
An adjective lacks the following paradigmatic forms:
(1) processive forms - the processive modifier -nun, the processive assertive -nunta/-nta, and the
processive adjunctive -nula;
(2) the intentive (-ulye/-lye) and purposive (-ule/-le) forms;
(3) subjunctive forms - the subjunctive attentive (= imperative) -ula, the subjunctive assertive (=
hortative or propositive) -ca, ... .
But the subjunctive assertive -ca occasionally occurs in the meaning ‘as soon as’ even with adjectives,
as in nal i ttattus haca ‘as soon as the weather is warm’, and in the meaning ‘as well as (being)’ even
with the copula (see ica in Part II). And adjective stems sometimes appear in the construction -e la,
normally expressing a plain command, but here carrying an exclamatory meaning ‘is indeed’: Ai
komawe la! ‘Heaven be praised! Thank goodness!’, Cham ulo take la! ‘How sweet it is!’, Ai koso
hay la! ‘Serves you right!’. Compare the copula form ila. It is unclear whether the bound particle la in
the two meanings of -e la is a single etymon or has separate origins, nor do we know whether it comes
from a reduction of the copula form in one or both of these usages, to say nothing of the imperative
ending -ula/-la itself. Moreover, a few adjectival-noun constructions appear as genuine commands (-e
la) and propositions (-ca), and in the intentive -ulye and purposive -ule forms. Acceptable examples
are given by CM (1:425, 428-9, 435) for chwungsil ‘being faithful’, co’.yong ‘being quiet’, taytam
‘being bold’, and minchep ‘being alert’. Co Sek.yen tells me she will accept chwungsil hay la,
chwungsil halye hanta, etc., but rejects such forms as chwungsil hale kanta ‘goes for the purpose of
being faithful’, a situation more processive in nature than ‘goes with the intention of being faithful’.
These few exceptions can simply be handled as semantic extensions or as abbreviations of some
processive structure (with the meaning of ‘acting, behaving’ rather than ‘being’).
The COPULA (cop) is a descriptive verb (an adjective) which almost never occurs after pause, for
it predicates a noun or noun phrase, with the meaning of ‘it is - ’, ‘it is a case (instance, kind) of - ’,
‘it is identified or specified as - ’, or (after a nominalization) ‘it’s that - ’. The stem is i- (defective
literary variant ilo-, defective formal variant iolssi-) after a consonant. After a vowel the initial III is
usually shortened to zero, i.e. dropped, when a consonant follows, as in pata ’ta = pata ita ‘it’s the
sea’; when a vowel follows, the stem vowel is reduced to y-, as in pata yess.ta = pata iess.ta ‘it was
the sea’, pata ye (se) = pata ie (se) ‘it’s the sea (and ... )’. In written Korean the i- after a vowel may
or may not be reduced or dropped, but a full vowel syllable is usually written in forms that would
otherwise leave a single consonant stranded: im (substantive), in (modifier), il(q) (prospective
modifier). Cf Mkk 1960:5:27. In addition to the various limitations of the adjective with respect to
inflectional endings, the copula also lacks the projective and adverbative forms: we find no occurrences
of *itolok ‘to the point where it is - ’ or *ikey ‘so that it is - ’, though Middle Korean had the
expected forms.
At least fourteen stems underlie complete paradigms as both processive and descriptive verbs:
cala- ‘reach, grow; be sufficient, enough’ mac- ‘tally; be correct’
etwuw- 1 ‘get dark; be dark’ mulu- ‘get soft; be soft’
ha- ‘do; be’ na(s)- ‘get better; be better’
huli- ‘get cloudy; be cloudy’ nuc- ‘get late; be late’
khu- ‘get big; be big’ nulk- 3 ‘get old; be old’
kiph- 2 ‘get deep; be deep’ palk- ‘get bright; be bright’
kwut- ‘get hard; be hard’ pulk- ‘get red; be red’
kyeysi- ‘exist; stay’ [honorific] telew- 1 ‘get dirty; be dirty’
1 Some speakers reject etwuw- and telew- as processive verbs. Examples: Nal i
etwup.nunta - ppallay lul ketwe la ‘It’s getting dark - gather the laundry up’; Keki
anc.ci mala - os i telep.nunta ‘Don’t sit there - you’ll get your clothes dirty’.
2 Apparently kiph- is processive only in pam i kiph.ess.ta ‘the night has deepened’,
perhaps by analogy with nuc.ess.ta ‘it has gotten late’.
3 For some speakers, at least.
218 PART I
A Reference Grammar of Korean
With less predictable differences in meaning, certain shapes represent both processive verbs and
adjectives: kolu- ‘is even’ or ‘makes it even’, pet- ‘(tooth) is protruding’ or ‘it stretches out; stretches
it out’, ... . (See Appendix 1.)
The intransitive verbs iss- ‘stay; exist, be located; have’, eps- ‘not exist; not have’, and tense
markers deriving from iss- (the past -ess- and the future -keyss-) have all the processive forms except
the processive assertive, but iss.nunta occurs in the meaning ‘stays’ with the negative forms mds(q)
iss.nunta and an iss.nunta (see §11.7.3). For a plain-style statement ‘exist, is (at); has’ you use iss.ta
not (*)iss.nunta and eps.ta not *eps.nunta. Yet in all meanings the processive modifier forms iss.nun
and eps.nun are more common than the simple modifiers iss.un and eps.un, which they have largely
replaced in standard speech. (But the simple modifiers are found in Phyengan; see Part II.) The
preceding description is inadequate for modern Seoul speech (*Yi Tongcay 1990), which differentiates
iss-i ‘stay’ from iss *2 (the other meanings) and treats it as a regular intransitive verb that shares most
but not all of its paradigmatic forms with the quasi-processive verb. To reconcile the two descriptions,
we might suggest that there are speakers, or at least writers, who replace iss.nunta ‘stays’ with the
shorter iss.ta (which also has the other meanings), perhaps obligatorily in certain situations, such as
quotations, and optionally in others, such as negatives. For "eps- the MK texts attest both the modifier
"ep'sun and (more often) the processive modifier "ep non; the predication can be negativized as "ep'ti
a'ni °ho-, but I have not found *a ni "eps-. For the stem isfi)- the modifier is isin, the prospective
modifier is isil(q); the processive modifier is is.non and its modulated form is is'nwon. The predication
can be negativized as is'ti a 'ni °ho-, but I have not found (l*)a ni is(i)-. Examples of the Middle
Korean forms will be found under their entries in Part II.
The past is simply -ess.ta and the future -keyss.ta, irrespective of whether the stem is processive or
descriptive. These stems and markers seem to lack any use of the subjunctive forms, the intentive, or
the purposive, except for iss- itself in the meaning ‘stay’, and they are not common even for it. These
verbs and bound auxiliaries (the tense markers) can be called quasi-processive in their behavior; we
will label them quasi-verbs intransitive (qvi). Ceng Insung 1960:262-3 neatly describes iss- as “a verb
that lacks the form *iss.nunta” (but that is not true of modern Seoul speech) and eps- as “an adjective
that has the extra form eps.nun”. He says that the honorific kyeysi- is like iss-, yet I find that both
kyeysita and kyeysinta are used for the plain present. The stem eps- occurs in the construction -e
cinta ‘gets to be’, normally limited to adjective stems though there are a few cases with intransitive
processive verbs (see the entry in Part II); *iss.e cinta does not occur. Notice that eps- has the derived
adverb form eps.i (< "ep'si) ‘without’, but (*)iss.i (*i'si) ‘with’ occurs only in dialects (p. 584).
The stem iss- is particularly tricky; see p. 319 for evidence that it should be treated as three
homonyms ‘stays’, ‘is’, and ‘has’. In the meaning ‘is’ and ‘has’, the appropriate auxiliary for such
nuclear focus conversions as Vpki nun ha- = V r ki nun V r is the descriptive hata (with the modifier
han and not hanun), but in the meaning ‘stays’ the processive hanta (with the modifier hanun) is
called for: Cip ey iss.ula ’ni-kka iss.ki nun hanta man un ku salam i olq ka? ‘Told to stay home, I
do stay home but will he come?’; Ku palam-cayngi ka cip ey kakkum iss.ki to hanun kes ul kkwum
ey to mollass.ta ‘I didn’t even dream that playboy would stay home every now and then, too’.
Japanese, on the other hand, uses the auxiliary suru (its counterpart of hanta) for a 1 r- ‘be’ and ‘have’
as well as for [wJo'V- or [w]i- ‘stay’: a' , ri/[wJo' 1 ri wa suru.
A causative verb (vc) is a transitive verb which is a member of a pair of stems that are related
in shape. The other member of the pair is active: either transitive or intransitive, and if intransitive
either processive or descriptive. The causative stem differs from the active stem by the suffixing of a
causative formative, which has several different shapes (§7.4). Such verbs are lexical causatives that
must be listed in the dictionary; they are not freely derived. But most verbs can freely participate in a
syntactic structure we will call the periphrastic causative (-key hanta, etc.). On similarities and
differences between these two kinds of causative, see §11.6.
A passive verb (vp) is a member of a shape-related pair, of which the active member is transitive
and the passive member is usually an intransitive processive verb, but some passives take objects (see
A Reference Grammar of Korean
PART I 219
§7.4, §11.6). The passive member is marked by a suffixed formative (§7.4). A CAUSATIVE verb is made
by attaching a similar formative to the stem of an active verb, which may be intransitive or transitive.
Reference here is to lexical passive and causative verbs; they are listed in dictionaries, but you do
not make up new ones. On the other hand, the periphrastic causative and passive structures are freely
created as needed. See §11.6 for the voice conversions.
An AUXILIARY verb (aux) is used in close juxtaposition with some other verb, which is usually in
the infinitive (-%) or gerund (-ko) form. The auxiliary conveys a somewhat different meaning from
what its stem carries when it stands as an independent verb, if the stem can so be used. There are
processive auxiliaries (= auxiliary verbs proper) and descriptive auxiliaries (= auxiliary adjectives).
Some of the auxiliaries are separable: a particle may intervene between the auxiliary and the verb
form with which it is used. If nothing can intervene, the auxiliary is inseparable.
A postnominal verb is a verb that is used in close juxtaposition with a verbal noun. Separable
postnominal verbs (ha- ‘be’; ha- ‘do’, ka- ‘go’, sikhi- ‘cause’, toy- ‘become’) are sometimes set apart
from the verbal noun by a particle; inseparable postnominal verbs (low- ‘be characterized by’,
sulew- ‘be, give the impression of being’, taw- ‘be like’; keli- or tay- ‘behave so as to give the
impression of’) occur only right after the noun.
7.2. Bound verbs.
Certain verbs are always attached to some other verb, or to a noun. Among these bound verbs,
we can single out defective infinitives, bound adjectives, and bound postverbs (or verb suffixes).
7.2.1. Defective infinitives.
Most of the defective infinitives, recognizable by the characteristic infinitive ending -el- a (§9.4),
often preceded by -(u)l- (which seems to be some sort of formative), are inseparably attached to the
auxiliary verbs ci- (vi) and ttuli- (vt):
ekule ‘dislocate, go against’ pekule ‘split, separate’
hay(e) ‘wear (out)’ pule ‘break’
hune ‘demolish; collapse’ pulke ‘bulge out’
kasule ‘bristle’ sakul% ‘collapse, wither’
(k)kopul% ‘bend’ sosul% ‘frighten, startle’
okule, wukule ‘curl up, warp; break’ ssule ‘topple’
pasule, pusule ‘break’ thute/ttut.e ‘tear’ [dialect] (= the < thu-)
But (k)kiwul.e (ci-/ttuli-) ‘tilt’ is the infinitive of an adjective (k)kiwu-l- ‘tilted, aslant’, and mune
(ci-/ttuli-) ‘demolish’ is the infinitive of the transitive verb mun(u)- ‘demolish’.
Five defective infinitives serve as verbal or adjectival nouns predicated by the postnominal
verb/adjective stem ha-:
ile ‘being/doing like this, so’ ette ‘being/doing what way, how’
kule ‘being/doing like that, so’ amule ‘being/?doing any way’
cele ‘being/doing like that, so’
The verbal use (‘doing’) for the first four was not present for the corresponding MK forms i'le, ku'le,
'tye'le, "es'te, and "a'mo'la (which is attested in very few forms), and it largely went unnoticed until
quite recently.
Only two defective infinitives seem to have other (also inseparable) attachments: kule ‘dragging,
pulling’ — apparently from an unattested variant of kkul.e ‘pulling’ (coming from a formation made on
the prospective modifier ku[z]ul(q) of MK kuz-), in kule mou- ‘rake up’, kule tangki- ‘gather and
puli’, kule cap- ‘grasp, clutch’, ... ; and wule ‘coming off and/or up’, which is probably an
abbreviation of wulele *- wulelu- ‘lift one’s head up’, in wule na- ‘soak out’, and wule na-o-
‘spring/well up’. We might want to include here certain elements of obscure or aberrant etymology
such as kala-(anc- ‘sink’); kalma-(tu-l- ‘alternate’); kelme-(ci- ‘shoulder, bear’); kelthe-(anc- ‘sit
astride’, tha- ‘mount astride’, tul.i- ‘bring it all in’, mek- ‘gobble it all up’); p%lke-(pes- ‘strip
naked’); cwuce-(anc- ‘slump, fall, cease’).
220 PART I
A Reference Grammar of Korean
The Hankul orthography treats none of these bound elements as infinitives, so it fails to set off
the -% ending in any of them, and our Romanization neglects the boundaries, too.
7.2.2. Bound adjectives.
Certain elements, usually treated by Korean grammarians as suffixes, derive adjectives from
inflected stems (and, in some cases, bound nouns). They are described in these entries of Part II:
-ta(h)-, -tala(h)- ‘be rather ••• ’
-a(h)-/-e(h)- ‘be rather ■■■ ’
-tama(h)-, abbreviation of -tama ha- (bound adjectival noun, §5.6.5)
-ep- (variants -up-, -ap-) / -p- = -ew- (-uw-, -aw-) / -w-
-pu- (derives adjectives from adjectives, verbs, bound nouns)
7.2.3. Bound postverbs.
By bound postverb we refer to certain formatives that are suffixed to verb stems to form new
stems, usually intensive, causative, or passive:
INTENSIVE -chi- -li-
-chwu- -lu-
-khi-
CAUSATIVE, PASSIVE -chwu- -li-
-hwu- (spelled -chwu-) -(i)wu-
-chi- -wu-
-i- -kwu-
-hi- -ay-
-ki- -y-
-ukhi- -u-
-ikhi-
For details and examples, see Part II and also §7.4, §11.6.
We might wish to consider as “bound postverbs”, too, the past and future markers (-ess- etc. and
-keyss-) that I have chosen to include with the endings for the following reason. These elements occur
only before some, but not all, of the final “mood” endings, unlike the other elements we are calling
bound postverbs. Moreover, the tense markers attach freely to all stems, unlike the bound postverbs,
for which the stems must be individually specified. It is this last criterion alone which excludes from
the bound postverbs the honorific marker -usi-, for it occurs freely before all the mood endings, as
well as before the tense markers. A final criterion would be simply one of sequence: whatever the
decision, we will see that the bound postverbs making intensives, causatives, and passives occur in a
position BEFORE the series of positions described for the verb endings in §9.1.
7.3. Defective verbs.
Several verbs occur in only a few of the paradigmatic forms:
ta-1- ‘request’ -» tao, tako (takwu), talla (notice that talla ’nta is an abbreviation of talla ko
hanta); on the irregularity of the forms, see Part II
tepu-1- ‘accompany’ -» tepulko, tepul.e (se)
kalo- ‘say’ -» kalotoy
( ••• ulo) malmiam- (§6.3) malmiam.% (se), malmiam.un
chamq ta(la)h- ‘be gentle’ -* chamq ta(la)h.key
Choy Hyenpay 1959:338-9 also includes a bound infinitive taka ‘(drawing) near’ from a supposed
*tak-, but there is no reason not to consider this to be just the expected literary or dialect variant
infinitive of taku- ‘bring (draw) close’, for which the normal colloquial infinitive in Seoul is take. To
be sure, taku- itself probably comes from *tak- (> tah- ‘touch, arrive, ... ’) + the causative postverb
-u-, but we can ignore the etymology in our synchronic description. To the contrary, Choy Hyenpay
1960 (Hankul 127:7-27). See also the note on -ta (ka) in Part II.
A Reference Grammar of Korean
PART I 221
Certain intransitive verbs or phrases with the meaning ‘gets to be’ occur only in the past or in
the simple modifier form, and these are sometimes mistaken for adjectives, since the usual translation
of ‘has gotten to be’ is ‘is’. Proof that they are not adjectives is that the past is necessary to give the
present resultative meaning: cal nass.ta ‘is nicely formed’, mos nass.ta ‘is ugly’, seylyen toyess.ta ‘is
refined’. A few expressions are treated, apparently in free variation, as either intransitive processiye
verbs or as adjectives: pala cyess.ta or pala cita ‘is shallow’; perhaps pyel nass.ta (? - rejected by
Co Sek.yen) or pyel nata ‘is odd’. On the other hand, there are some cases which appear only as
adjectives: emcheng nata ‘is enormous’, mo nata ‘is angular = is difficult in personality’, ttwukpyel
nata ‘is quick-tempered’. KEd treats hoth ci- and hwumi ci- as the intransitive ‘become simple’ and
‘form a bend; get deep/secluded’, but both NKd and LHS cite them as adjectives.
The variant copula stems ilo- and iolssi- are also defective; see Part II. I am not quite sure why
LHS lists mata- ‘reject, abhor’ as a defective transitive verb. The entry in Kim Minswu and Hong
Wungsen’s dictionary has mata- simply “transitive”. NKd does not list the verb, which might be
thought to come from a contraction of *ma ha-ta with the infinitive of the MK adjective "ma-
‘dislikable, disliked’, a structure equivalent to modern silh.e hanta ‘dislikes’. But the lone MK
example is cay-"myey lol "mata khesi'nol ‘though he treats the alms rice as despised’ (1459 Wei
8:78a). If the structure is parallel to the -e hanta of modern Korean, then LCT was wrong in his
citation form “ "ma 'ta for he should have used “ "mata ta . If he took the structure to be «- "ma-[']ta
ho- kesi nol ‘though he says/feels it is despicable’, the entry citation is correct (but there is something
odd about the accent). In that case, the transitive stem mata- derives from a contraction of the
quotative structure -ta (ha)-. It is not clear just when the use of hanta to transitivize the adjective
infinitive developed; it may be not be all that old. The only other use of "ma- cited by LCT is ku lol
mata kwos nekisimyen ‘if you regard it precisely as despicable’ (1676 Sin.e 4:15a). Perhaps all these
uses are semantic extensions of the verb “maflj'ta ‘desists’, i.e. “rejects”.
7.4. Causative and passive verbs.
There are related pairs of Korean verbs which differ in what is called voice. We find two major
types: passive related to active, and causative related to active. The relationship is both semantic
and formal. The underlying stem is usually the same, but the passive or causative includes a bound
postverb (§7.2.3). Causative verbs (vc) are always transitive, and passive verbs (vp) are typically
intransitive (= vpi) but some take a few objects so we will label them transitive passive (= vpt). The
types of derivation can be seen from the following scheme:
VC
•*-
vt
«-
vi
■*-
adj
vp(i)
«-
vt
vpt
■*-
vt
One unusual case looks like a vp from a vi: menci ka palam ey pullinta ‘the dust gets blown by
the wind’, Cf palam i punta ‘the wind blows’ and the lack of a *palam i menci lul punta ‘the wind
blows the dust’. But the most appropriate derivation appears to be something like this: (salam i) menci
lul punta ‘(a person [etc.]) blows the dust’ -*■ menci ka (salam eykey) pullinta ‘the dust gets blown
(by the person)’ -*■ [palam ey (iss.ta) ‘(it is) in the wind’ -* ] menci ka palam ey pullinta ‘the dust
gets blown in the wind = it blows in the wind’. A couple of similar cases are kamki (ey) tullinta
‘catches a cold’ (= kamki ka tunta), tali lul tachinta ‘gets injured on the leg’ from tah.chi- from
tah- ‘touch’ (vi). (Most of the examples given by CM 1:273, however, seem to be in error.)
The transitive passives all seem to refer to unsought suffering:
cap.hi- ‘have someone take (from one)’ < cap- ‘take’: pal ul cap.hinta ‘gets caught by the leg’,
kyelqcem ul cap.hinta ‘has fault found with one’,
ccalli- ‘get it cut’ < ccalu- ‘cut’: mok ul ccallinta ‘gets one’s throat cut = gets fired’,
ccilli- ‘get it pierced (stabbed, pricked)’ < ccilu- ‘stab’.
222 PART I
A Reference Grammar of Korean
chayi- ‘get kicked on /in’ < chay- ‘kick’: kasum ul chayinta ‘gets kicked in the chest’ (compare
kong i cal chayinta ‘the ball kicks well’),
cilli- ‘get kicked in the ••• ’ < cilu- ‘kick’,
halth.i- /halchi-/ ‘get it licked or swiped’ < halth- ‘lick’.
kkayi- ‘get hit on’ [dialect? - CF NKd 4231a] < kkay- ‘hit’: cengkangi lul kkayinta ‘gets hit in
the shin’.
kka(y)kk.i- ‘get it scraped’ < kkakk- ‘scrape’: salam i nach ul kka(y)kk.inta ‘a person gets his
face scraped = loses face’ - and that can also be said as salam i nach i kka(y)kk.inta ‘a person,
his face gets scraped’.
mek.hi- ‘have it eaten up’: talun salam hanthey ton ul mek.hinta ‘has one’s money “eaten up” =
gets swindled by another person’.
peyi- ‘get cut on’ < pey- ‘cut’: sonq-kalak ul khal ey peyyess.ta ‘got one’s finger cut on a knife’,
ppays.ki- ‘have someone grab, suffer the loss of’ < ppays- < ppay-as- ‘grab’: kongsan kwun
hanthey Sanghay lul ppays.kyess.ta ‘suffered the loss of Shanghai to the Communist troops’,
ttelli- ‘get stripped (= robbed) of’ < tte-1- ‘shake off’: ton ul ttellinta ‘is stripped of one’s
money’.
According to Song Sekcwung 1967:177, the transitive passive will occur only when there is a WHOLE-
PART relationship between the two nouns. Occasionally a causative is used with unexpected meanings
similar to those of the transitive passive: seykan ul mocoli thaywunta ‘(burns up =) loses all one’s
furniture in a fire’.
The basic meaning of most causative verbs is something like ‘makes it so that something happens
or is’. There is no good general translation in English, which offers a choice of ‘makes someone do it’,
suggesting force and coercion, or ‘lets someone do it’, suggesting permission to do something the other
person wants to do. The expressions ‘has someone do it’ and ‘gets someone to do it’ are ambiguous,
for they can translate causatives, passives, or simple favors. The Korean causative implies neither force
nor permission. Only from the context can you tell whether the act someone is caused to perform is
something he wants to do, or something he is forced to do.
The basic meaning of most passive verbs is something like ‘gets so that something happens to it or
might happen to it’. Often the meaning of a Korean passive verb has some extra flavor of available
or potential undergoing of an action. Compare san i pointa ‘the mountains can be (are available to
be) seen - whether anyone is looking or not’ with san ul ponta ‘the mountains are seen; someone sees
the mountains’. Similar examples (CM 1:328) are yeki se nun koki ka cal cap.hinta ‘a lot of fish are
caught here’ and i mun un swipkey yellinta ‘this door opens easily (is easily opened)’. In English we
often use passives to avoid committing ourselves to the identification of the subject. In Korean the
device is unnecessary for that purpose, since an overt subject is not required by Korean sentences.
For more on the syntax of causatives and passives, see §11.6.
The morpheme for the causative bound postverb has several shapes which are identical with the
shapes of the passive. As a result, convergence sometimes produces homonymous causative and
passive forms from the same active stem:
anki- vp ‘get embraced’, vc ‘embrace’ *- an- vt ‘embrace’
cap.hi- vp ‘get caught’, vc ‘cause to catch’ *- cap- vt ‘catch’
ep.hi- vp ‘get carried’, vc ‘cause to carry’ *- ep- vt ‘carry (on back)’
halth.i- /halchi-/ vp ‘get licked’, vc ‘cause to lick’ *- vt halth- ‘lick’
ilk.hi- vp ‘get read’, vc ‘cause to read’ *- ilk- vt ‘read’
kkakk.i- vp ‘be cut’, vc ‘cause to cut’ *- kkakk- vt ‘cut’
kwup.hi- vp ‘get broiled’, vc ‘make someone broil’ *- vt kwuw- ‘broil’
mulli- vp ‘get bitten’, vc ‘cause to bite’ *- mu-1- vt ‘bite’
palp.hi- vp ‘get stepped on’, vc ‘cause to step on’ *- vt palp- ‘step on’
poi- (poy-) vp ‘be seen’, vc ‘cause to see, show’ *- vt po- ‘see’
A Reference Grammar of Korean
PART I 223
silli- vp ‘get loaded’, vc ‘cause to load’ *- sit-/sil- vt ‘load, carry’
simki- vp ‘get planted’, vc ‘cause to plant’ *- sim- vt ‘plant’
ssip.hi- vp ‘get chewed’, vc ‘cause to chew’ *- ssip- vt ‘chew’
ssui- vp ‘get used; get written’, vc ‘cause to use; cause to write’ *- ssu- ‘use; write’
ttut.ki- vp ‘get bitten’, vc ‘cause to graze’ ttut- ‘bite, graze’
tulli- vp ‘get heard’, vc ‘cause to hear’ *- tut-/ tul- ‘hear’
These homonymous pairs of causatives and passives are not differentiated by accent or vowel length in
standard speech, despite what is said in CM 1:275-6. Cf Ceng Insung 1960:92-3.
In certain stems we find a string that could be representing one of these two formatives, but either
there is no underlying active counterpart to justify an analysis like sik-hi- for sikhi- ‘cause’ (can we
perhaps analyze the stem as s-ikhi- and take s- to be an allomorph of ha-?); or, if there is an
underlying form (tul-li- = tul-lu- ‘drop in’, tu-l- ‘enter’) the syntactic relationship between the two
does not correspond to that expected for voice-related pairs. Verbs of this sort lead us to recognize
another derivative bound postverb that happens to be of the same shape as the causative. It might be
appropriate to treat some of the odd cases of passives formed on intransitives in the same way. There
is also tuli- ‘give to a superior’, which is historically a causative formed on tu-l- ‘hold up’.
The passive and causative bound postverbs have so many shapes in common that we will consider
them together. The shapes divide into two thematic groups: (1) those which include the phoneme i,
and (2) those which include the phoneme wu (but not the phoneme i); there is also (3) an athematic
group, in which we will include the reductions of i to y. Formatives with the wu theme (and,
apparently, the athematic formatives) all form only causatives, with the possible exception of one
questionable stem, puliwu- vp ‘be employed’ < puli- vt ‘employ’. Formatives with the i-theme form
both causatives and passives.
The shapes are listed below with critical examples given for each shape, c = causative, p =
passive; T = transitive, I = intransitive, A = adjective (descriptive stem). The down arrow (*) means
the preceding string from the stem is omitted (deleted) when the formative is added, and the right
arrow (-*) means the stem string on the left is replaced by the string on the right; these are synchronic
statements, not necessarily recapitulating the history of the formations. Notice that the two distinct
origins of the modern — lu-/ — 11- stems (§8.3.1) are reflected in the derived stems: hulu-/hull- ‘flow’
makes the causative hulli- < hul'l-i- ‘make flow’, and olu-/oll- ‘rise’ makes the causative olli- <
wol'G-i- ‘raise’.
1. i theme (causative, passive)
-i- cwuk.i- ‘kill’ (c) cwuk- ‘die’ (I)
mek.i- ‘feed’ (c) *- mek- ‘eat’ (T)
kiwul.i- ‘tilt’ (c) <- kiwu-1- ‘be tilted’ (a)
noph.i- ‘heighten’ (c) *- noph- ‘be high’ (a)
noh.i- ‘get put’ (p) *- noh- ‘put’ (t)
nanw(u)i- = nanwe ci- ‘be divided’ (p) nanwu- ‘divide’ (T)
hulli- ‘make flow’ (C) *- hulu-/hull- ‘flow’ (I)
-hi- anc.hi- ‘seat’ (c) *- anc- ‘sit down’ (i)
kkulh.i- ‘makes it boil’ (C) *- kkulh- ‘boil’ (I)
kwup.hi- ‘bend’ (c) *- kwup- ‘be bent’ (a)
ip.hi- ‘cause to wear’ (c) *- ip- ‘wear’ (t)
mek.hi- ‘get eaten’ (p) *- mek- ‘eat’ (T)
(aw^-i- akki- ‘spare; value’ (c) akkaw- ‘is regretful; is precious’ (a)
(wu'I)-hi- kat.hi- ‘be confined’ (P) *- katwu- ‘confine’ (T)
-ki- swumki- ‘conceal’ (c) swum- ‘be hidden’ (I)
olm.ki- ‘move it’ (c) *- olm- ‘move’ (i)
pes.ki- ‘unclothe’ (c) *- pes- ‘take off, remove’ (t)
ccic.ki- ‘get torn’ (p) *- ccic- ‘tear it’ (T)
224 PART I
A Reference Grammar of Korean
-chi- kuluchi- ‘ruin’ (c) *- kulu- ‘be wrong’ (a)
sos.chi- ‘raise; exasperate’ (c) «- sos- ‘tower up, rise’ (i)
- But /sochi-/ could be treated as sos-hi-.
-ukhi- il.ukhi- ‘raise’ (c) i-1- ‘rise’ (I)
-Ikhi- tol.ikhi- ‘turn (head)’ (c) <- td-1- ‘turn’ (I)
—I-li- < -Gi- alii- ‘inform’ (c) a-1- ‘know’ (t)
salli- ‘let live’ (c) «- sa-1- ‘live’ (I)
mongkulli- ‘make (grain) awnless’ (c) *- mongku-1- ‘be awnless’ (a)
phalli- ‘get sold’ (p) *- pha-1- ‘sell’ (T)
olli- ‘raise’ (c) *- olu-/oIl- ‘rise’ (i)
Almost unique: pemulli- ‘be mixed’ (P) and ‘cause to mix’ (c) pemuli- ‘mix’ (t). The form pemuli-
was earlier pe muli-, a causative *- pe 'mul- ‘whirl it (around)’ (t) and ‘tie it up’ (T), surviving
figuratively in pemu-l- ‘be involved, mixed up in’. Apparently similar: holli- ‘get infatuated’ (P) *-
holi- ‘infatuate’ (T). And killi- ‘get raised’ (< *kil'Gi- < *ki'l[u]-G-i-) is a late development.
2. wu theme (causative only)
-wu- kkaywu- ‘wake someone’ (c) *- kkay- ‘come awake’ (i)
kel.wu- = ‘fertilize’ (c) «- ke-1- ‘be fertile’
tot.wu- ‘raise it’ (c) «- tot- ‘rise’ (I)
(u*)-wu- palwu- ‘straighten’ (c) [usually replaced by palo cap-] *- palu- ‘be right’ (A)
-chwu- kot.chwu- ‘straighten’ (c) *- kot- ‘be straight’
-(c)hwu- nac.chwu- ‘abase, make low’ (c) *- nac- ‘be low’ (a)
nuc.chwu- ‘loosen; delay’ (c) *- nuc- ‘be slack; be late’ (a),
mac.chwu- ‘spell’ (c) *- mac- ‘be correct’ (A)
cac.chwu- ‘quicken’ (c) *- cac- ‘be incessant’ (a)
kac.chwu- ‘prepare’ (c) «- kac- = kaci- ‘have’ (t)
(u-*ey)-wu- seywu- ‘make stand, establish’ (c) «- su- ‘stand’ (I)
(ew-*ey)-wu- teywu- ‘heat’ (c) *- tew- ‘be hot’ (a)
-ywu- < -i-wu- caywu- ‘put to sleep’ (c) *- ca- ‘sleep’ (i)
-kwu- sos.kwu- ‘make rise’ (c) *- sos- ‘spring up’ (I)
tot.kwu- ‘make it higher’, (= tot.wu-) ‘raise it’ (c) *- tot- ‘rise’ (I)
3. athematic (causative, passive)
-ay- eps.ay- ‘eliminate; use up’ (c) *- eps- ‘be nonexistent’ (i) - but see p. 429!
-y- nay- ‘put out’ (c) *- na- ‘emerge’ (I)
cay- = caywu- ‘put to sleep’ (c) *- ca- ‘sleep’ (i)
poy- = poi- ‘show, let see’ (c), ‘get seen’ (p) *- po- ‘see’ (T)
(ul)-ey- kenney- ‘carry over’ (c) *- kennu- ‘cross over’ (t)
(h'l)-y- tay- ‘bring in contact’ (c) *- tah- ‘come in contact’ (I)
-u- kilu- ‘raise’ (c) *- ki-1- ‘get big’ (I) - see below
The voice-deriving bound postverbs should not be confused with the intensive bound postverb
-chi-, which is morphemically related to the auxiliary verb chi- that is used after the infinitive as an
intensifier. See Part II for examples of the postverbs, and compare Choy Hyenpay 1959:351. The
intensive postverb seems to have the shape -chwu- in tulchwu- ‘raise, expose, ... ’ {*- tu-1- ‘hold up;
lift; cite’), but there is a dialect variant with the expected shape tulchi-, and there may be some
connection with the obsolete verb chi- ‘raise’ (Cf the bound preverb tul-); there is also tulkhi- ‘get
discovered, caught’, a specialized use of a passive from the same tu-1-. I have not included -chwu-
among the shapes of the intensive, but the -li-/-lu- of tulli- / tullu- ‘drop in’ (*- tu-1- ‘enter’) is there.
The -khi- of tulkhi- is not listed, only the causative -ikhi- (probably < -i-khi-) and -ukhi-.
Some of the complications of shape are the result of phonological changes from earlier forms. The
Middle Korean causatives and passives were made with these formatives:
A Reference Grammar of Korean
PART I 225
- 1 -
-C)y-
- ki-
-hi-
-Gi-
-Gw u /o-
-Gw u /oy-
< *-G-
cwu'ki- ‘kill’ (c) *- cwuk- ‘die’ (i)
me 'ki- ‘feed’ (c) *- mek- ‘eat’ (t)
non'hwoi- ‘get divided’ (p) *- non'hwo- ‘divide’ (t)
na'hi- ‘cause to give birth to’ (c) *- nah- ‘bear’ (t)
il hi- ‘cause to lose’ (c) *- ilh- ‘lose’ (t)
hul'li- ‘make flow’ (c) *- hulu-lhull- ‘flow’ (i)
wol'Gi- ‘raise’ (c) *- wolo-lwolG- ‘rise’ (i)
"cay- ‘cause to sleep’ (c) *- °ca- ‘sleep’ (i)
"nay- ‘cause to emerge’ (c) *- °na- ‘emerge’ (i)
"ken'ney- ‘carry it over’ (c) *- "ken'ne- ‘cross over’ (i)
"pwoy- ‘cause to see’ (c), ‘gets seen’ (p) *- °pwo- ‘see’ (t)
"syey- ‘cause to stand’ (c) *- °sye- ‘stand’ (i)
swum 'ki- ‘conceal’ (c) *- 'swum- ‘lie in hiding’ (i)
wolm ki- ‘move it’ (c) *- "wolm- ‘move’ (i)
pes'ki- ‘unclothe’ (c) *- pes- ‘remove, strip off’ (t)
an ' chi- ‘cause to sit’ (c) *- anc- ‘sit’ (i)
ca 'phi- ‘cause to catch’ (c), ‘get caught’ (p) < cap- ‘catch’ (t)
el'khi-, elk'khi-, elq'khi- = el'khi- ‘get tied’ (p) < elk- ‘tie’ (i)
me'khi-, mek 'hi-, mek'khi- ‘get eaten’ (p) < mek- ‘eat’ (t)
ni'phi-, nip'hi- ‘cause to wear’ (c) < nip- ‘wear’ (t)
pa 'khi-, pak'hi- ‘get stuck/printed’ (p) < pak- ‘stick in/on, print’ (t)
pat'hi- ‘cause to get’ (c) < pat- ‘get’ (t)
twot'hi- ‘cause to rise, raise’ (c) < twot- (i)
moyngkol'Gi- ‘cause to make’ (c) *- moyng'kol- ‘make’ (t)
mul'Gi- ‘get bitten’ (p) *- mul- ‘bite’ (t)
nol'Gi- ‘cause to fly’ (c) *- nol- ‘fly’ (i)
sal 'Gi- ‘cause to live’ (c) *- "sal- ‘live’ (i)
skol'Gi- ‘get spread’ (p) *- 'skol- ‘spread’ (t)
tul'Gi- ‘cause: to hear, to enter, to lift’ (c)
kil 'Gi- ‘increase it’ (c) *- "kil- ‘get big’ (i)
twol'Gi- ‘cause to turn’ (c) = twolo- *- "twol- ‘turn’ (i)
el'Gwu- ‘cause to freeze’ (c) *- "el- ‘freeze’ (i)
me'mulGwu-, me'mulGwo- ‘cause to stay’ (c) *- me'mul- ‘stay’ (i)
mey'Gwu- ‘cause to shoulder’ (c) *- "mey- ‘shoulder’ (t)
sul'Gwu- ‘cause to vanish’ (c) *- 'sul- ‘vanish’ (i)
tu li Gwu-, tu 'li Gwo- ‘cause to hang down’ (c) = tu li- (t)
twol'Gwo- ‘cause to turn’ (c) = twolo- / twolG- "twol- ‘turn’ (i)
kil'Gwu-, kil'Gwo-, kil'Gwuy- ‘raise’ (c) = kil %- *- "kil- ‘get big’ (i)
"nilGwuy- ‘cause to reach’ (c) *- ni'lu- = ni'lul- ‘reach’ (i)
al'Gwoy- ‘inform’ (c) *- "al- ‘know’ (t)
kil %-/ kilG- < *kil u /o-G- ‘raise’ (c) *- "kil- ‘get big’ (i)
twolo-1 twolG- < *twolo-G- ‘cause to turn’ (c) *- "twol- ‘turn’ (i)
The vowel of - Gw u /o- may represent an incorporation of the modulator - 'w%-; if so, the etymon of
the formative was simply -G-, as indicated also by < *-G-. The infinitives of the last two
examples are kil'Ge (1465 Wen 1:1:1:111a) and tol'Ga (1447 Sek 6:4b); Cf kil'Gwe (1447 Sek 9:17a)
and twol'Gwa (1632 Twusi-cwung 16:56b) = twol'Ga (1481 Twusi 16:55b), twol'Gye ( ? 1517- Pak
1:21b) - from twol-'Gi- (> tolli-). But "ep.si'Wa (1449 Kok 155), "ep.si'Wozo'Wa (1459 Wei
17:77a), and "ep'siWo'n i (1459 Wei 23:65b) indicate a labial *-i-p u /o- (echoing the p of "eps-1).
226 PART I
A Reference Grammar of Korean
7.5. Auxiliary verbs.
Auxiliary verbs are used in construction with preceding verb forms, most often the infinitive -e or
the gerund -ko; less often the suspective -ci, the adverbative -key, the transferentive -ta, and the
unusual cases of 34, 45, and 46 in the lists below. In these lists all the auxiliary descriptive verbs (=
auxiliary adjectives) are segregated from the auxiliary processive verbs (= auxiliary verbs proper), and
each auxiliary is preceded by an indication of the category of verb form with which it enters into a
construction: the letters A (adjective), C (copula), and V (processive verb) represent the stem before
the that designates the ending. With some misgivings I offer tag translations for the auxiliaries,
but the reader is urged to look up the separate entries in Part II for more detailed descriptions and
examples. The separable auxiliaries are marked by the symbol # placed after the number; they can be
set apart from the preceding verb form by at least the particles un/nun or tul. Since we distinguish
auxiliary constructions from simple compounds or sequences in which both verbs retain their usual
meanings and functions, not included are such forms as -e neh- ‘do and put in’. And idiomatic
formations such as ka(-)tah- ‘arrive (there)’ and wa(-)tah- ‘arrive (here)’, tay(-)ka- ‘arrive (there) in
time’ and tay(-)o- ‘arrive (here) in time’ are best put in the dictionary as lexical entries, since you will
seldom if ever separate the infinitive from the following stem, even by the particle se. At the left of
each item in the lists below there is information on attaching the honorific marker to the main verb (a),
to the auxiliary (/?), or to both (aft). A plus (+) means it is possible to attach -usi-, a minus means that
it is not possible. When the plus is italicized {+) the honorific marker is attached only under special
circumstances of one kind or another; even in some of the cases marked with the minus it is possible to
get the honorific verbs capswusi- ‘eat’, cwumusi- ‘sleep’, and kyeysi- ‘stay’ (also tol.a kasi- when it
means ‘die’), though not the regularly derived forms such as hasi- ‘deign to do’, kasi- ‘deign to go’,
and pat.usi- ‘deign to receive’.
Lists of auxiliary verbs
Auxiliary processive verbs ( aux v)
+ -usi-
a
§_
-
+
-
1.
n
V-e tay-
[intensive] ‘do hard / continuously’
-
+
-
2.
V-e chi-
[intensive] ‘do hard / continuously’
+
-
-
3.
V/A-e ssah-
‘do /be more than an ample extent’
-
-
-
4.
V/A-e ppa ci-
‘(get old, rotten, musty) through and through’
-
+
-
5.
V-e (p)peli-
[exhaustive] ‘do completely’
-
+
-
6.
#
V-e twu-
[completive] ‘get it done’
-
+
-
7.
n
V/A-e noh-
[anticipatory] ‘do for now / later’ / ‘be all - ’
+
-
-
8.
#
V-ko na-
[transitional] ‘just did, come from doing’
+
+
+
9.
#
V-e na-
[continuative] ‘keep doing, do and do again’ -
Cf -e na- as regular compound (kkay na-
‘recover one’s senses, come to’)
+
+
+
10.
#
V-ko ma-1-
[terminative] ‘finish(ed) doing’
?
?
?
V-ko ya ma-1-
‘end in doing, end up doing’
+
-
+
V/A/C-ko malko
‘of course do/be’
+
+
+
11.
#
V-ta (ka) ma-1-
‘do a while and then stop’
+
+
+
11a.
tt
V-ta (ka) mos ha-
‘fail to do, try but cannot’
+
+
+
12.
n
V-ko tani-
‘go around doing’ (Treat as regular compounds?)
-
+
-
13.
V-e mek-
[vulgar and pejorative] - Cf 44.
-
+
-
14.
V-e nay-
[perseverative] ‘do all the way’
-
+
-
15.
#
V/A-e ka-
[out-directive, exo-developmental] ‘away,
ongoing’ - Cf regular -ko ka- ‘go ... doing’
A Reference Grammar of Korean
PART I 227
—
+
-
16.
V/A-e o-
[in-directive, endo-developmental] ‘this way,
upcoming’ - Cf regular -ko o- ‘do and come’
+
+
+
17.
V-e po-
[exploratory] ‘try doing’ (not ‘try to do’ =
-ulye ko ha- or -ki ey him ssu-)
?
+
?
18.
A-e/-key poi- (poy-)
[semblative] ‘look like (it is)’
—
+
-
19.
V-e cwu-
[favor] ‘do for’ - separable by ’ta
+
-
-
19a.
V-e talla;
[reflexive request, §11.2] ‘ask someone to do for
V-e tao (tako)
for one’ - separable by ’ta
—
+
-
20.
V-e tuli-
[honorific favor] ‘do for (an esteemed person)’
—
+
-
21.
V-e pachi-
[honorific favor] ‘do for (an esteemed person)’
+
+
+
22.
V-ci anh-,
V-ci an(i) ha-
[negative] ‘not do’ - Cf 43.
+
+
+
23.
V-ci ma-1-
[prohibitive] ‘refrain from doing’ - usually in
a subjunctive form (imperative, hortative)
24.
(V) ha-
‘do’ - dummy verb (general auxiliary) in such
structures as V-ki to ha- ‘indeed/also do’
+
+
+
25.
V/A-key ha-
[causative] ‘make it so that’ - see §7.4
+
+
+
26.
V/A-key toy-
[externally conditioned gradual inceptive] ‘get
so that, come to do/be’
-
+
-
27.
A/V-key kwu-1-,
[behavioral] ‘act in a manner that is’ - usually
A-i kwu-1-
A-key, but mos kyentikey ‘unbearably’ *- vt
-
+
-
28.
A-e ha-
[cathecticizer, emotion transitivizer] ‘project
an emotion (toward something /someone)’
29.
V-e cilu-
= 30.
+
—
30.
V-e ttuli-
(1) turns vi into vt
(2) intensifies transitivity of vt and defective
infinitives
-
+
-
31.
V-e ci-
(1) [inceptive] ‘get to be, become, grow’ -
usually A-e ci- but sometimes V-e ci-, e.g.
pel.e ci- ‘split’ *- pe-1- vi
(2) ‘get/be done’: turns vt or defective inf into vi
(3) intensifies vi
+
+
+
32.
V/A-e (se)
cwuk.keyss.ta
‘so much one could die; very much’
+
+
+
32a.
V-e (se) mos
salkeyss.ta
‘so much one will not live; extremely’
+
+
+
32b.
V-e (se)
hon nass.ta
‘so much it frightened one witless; extremely’
+
-
-
33.
V/A-e kaciko
‘with (the accomplishment or resultant fact)’
+
-
-
34.
V-ulye ko tu-1-
‘threaten (try, be about) to do’
-
+
-
35.
V-e tu-1- / tul.i-
‘do into, upon, at’ (Treat as regular compounds?)
+
+
+
36.
V-ko iss-
[progressive] ‘be doing, be -ing’
—
+
—
37.
V-e iss-
[resultative] ‘be done, be -ed’
Auxiliary adjectives (aux adj]
+
+
+
38.
# V-ko siph-
[desiderative] ‘want to do’
-
-
-
39.
V-ko ci-
[literary] = 38.
-
-
-
V-ko ca/ce/cie
‘wanting/intending to do’ (= -ulye)
-
-
-
40.
V-e cii-
[literary] = 38.
228
PART I
A Reference Grammar of Korean
- 41.
tt
V-e ci-
+ -
- 42.
tt
(A) ha-
+ +
+ 43.
tt
A-ci anh-,
A-ci an(i) ha-
4- -
- 44.
A-e mek-
+ -
- 45.
tt
V-na po-,
V-nun ka po-
A-un ka po-
+ -
- 46.
tt
V/A/C-na siph-
-ta siph-
-ulila siph-
-ess.umyen siph-
in mes-tul.e ci- ‘be nice’; and in free
variation with 31 (aux. v.) in pala ci-
‘be (or get) shallow’ 1
‘be’ - dummy adjective (general auxiliary) in such
structures as A-ki to ha- ‘indeed/also be’
[negative] ‘not be’ - Cf 22
[vulgar and derogatory] - Cf 13
[semblative] look as though, seem
[semblative] feel as if - also postnominal verb
with tus, sang/seng, ka, ya, -ulq kes man
1 The critical examples are ku kes Cham mes-tul.e cita (not *cyess.ta) ‘that is real nice’ and
ku kes i pala cita/cyess.ta ‘that is shallow’.
7.6. Postnominal verbs.
Postnominal verbs occur in construction with preceding nouns, typically verbal nouns (§5.6).
Some of the postnominal verbs are separable, at least by the particles un/nun, to, or tul, and they are
marked with
Lists of postnominal verbs
Postnominal processive verbs (postnom v)
1.
keli-
‘behave in a way that creates the impression of; XX ha-
‘repeatedly/continuously do’
2.
tay-
[colloquial] = 1.
3.
k(h)uli-
‘do’
4.
ha-
‘behave, go (boom!, ... )’: X ha-, XX ha- = X keli-
5.
i-
‘behave, go’ - as in wumcik i- ‘move, budge’
5a.
na-
‘behave, feel’ (in kamcil na- ‘feel impatient’)
6.
tt
ha-
‘do’
7.
tt
tte-l-
‘do’
8.
tt
puli-
‘do’
9.
tt
phi(wu)-
‘do’
10.
tt
ppay-
‘do’
11.
tt
po-
‘do; see to’
12.
tt
(na-)ka-
‘go (out) to do’
13.
tt
(na-)o-
‘come (out) to do’
14.
tt
sikhi-
‘cause to do’ [causative] - see §7.4
15.
tt
toy-
‘get/be done’ [passive] - see §7.4
16.
tt
pat-
‘incur the doing of, have done to one’ [passive] - see §7.
? 17.
tt
ssu-
‘use, do; [? causative]’
18.
ci-
‘get/become characterized by’
19.
tt
chi-
‘do’
?20.
tt
cl(s)-
(= 6) in kyelqceng ci(s)-/ha- ‘decide’
21.
eps-
‘lack’ - can be treated as regular compounds
A Reference Grammar of Korean
PARTI 229
Postnominal adjectives (postnom adj)
?22.
eps-
‘be lacking/deficient in; be bad with respect to’: See Part II.
22a.
(#)
kwuc-
‘be bad with respect to’ (simswul kwuc- rarely is simswul i kwuc-)
23.
na-
‘be’ - pyel nata or pyel nass.ta [defective vi] ‘is special’;
mas nata/nanta [A/V] ‘is tasty’
24.
#
ha-
‘be’
25.
ha-
‘be, give the impression of being’: X ha-, XX ha-
26.
mac-
= 25
27.
sulew-
= 25
28.
(h)ow-
[obsolete] = 27 = 25
29.
low-
‘be characterized by’ [occurs only after vowel]
30.
( •q) toy-
‘be’
30a.
kath-
‘be ...-like’ - inseparable in engtheli kath-, ...
30b.
cha-
‘be full of / in’ - inseparable in an cha-, ...
31.
taw-
‘be like, be worthy of being’
32.
ci-, chi-
‘be, be characterized by’
32a.
cew-
‘be characterized by’ - Are the only examples swus cew-
‘be simple-hearted’ and pich cew- ‘be dignified’?
33.
#
ccek-
‘feel, give/have a feeling of’
34.
#
(tus) siph-
‘feel/look (like)’ - separable by to
35.
(s%ng) siph-
‘seem to be’ - separable by to
36.
(ka, ya, -ulq kes
man) siph-
‘feel/look (like)’
37.
(seng) p a / u lu-
[dialect] = 35
We can add to the list of postnominal verbs lop.hi- ‘makes it be characterized by’ (occurring only
after a vowel), a causative from the postnominal adjective low- (item 29 above). 1 have excluded from
the list coh- ‘be good with respect to, have a nice - ’, sanaw- ‘be bad with respect to, have a bad - ’,
and other such words carried in the lists of CM 1:447-9, because these adjectives seem quite freely
separable by the particle i/ka and therefore do not differ in kind from other adjectives that are
preceded by a noun (4- particle).
The stem ha- ‘do/say/be’, which has so many uses (see the entries in Part II), is both processive
and descriptive. The two uses contrast nicely in te hata (adj-n) ‘be more, be worse’ and te hanta (vni)
‘get worse (exaggerated)’ or (vnt.) ‘add, gain, increase’.
7.7. Recursiveness of auxiliary conversions.
What, if anything, constrains the application of an auxiliary conversion to a sentence that had
already undergone an auxiliary conversion? Apparently the only restrictions are those of semantic
incompatibility (the result does not make sense), or unwieldiness (the sentence is too heavily burdened
for easy processing). If it were not for these constraints, an infinite number of longer and longer
sentences could be created just by reapplying auxiliary conversions. There are a fair number of
sentence types with a sequence of no more than two auxiliary conversions. The table below, compiled
by J Yi Tongcay in 1962, shows those sequences of two infinitive-auxiliary conversions that are
acceptable (+) and those that are not (-). The auxiliaries are assigned letters (A to X) arbitrarily and
these are used as references at the left (the first conversion) and at the top (the second conversion).
The key to the letters is on the extreme left. If you want to know whether it is permissible to reapply -e
noh- to a sentence which has already had it, you look for the “G-G” intersection and find a that
means there are no such sentences as *hay noh.a noh.ass.ta ‘did it for later for later’, though it is
230 PART I
A Reference Grammar of Korean
possible to apply the conversion to a main verb noh- ‘put’ as in phyenci lul chayk-sang wi ey noh.a
noh.ass.ta ‘I put the letter on the desk for the time being’. An examination of the table will reveal that
in general a conversion cannot be reapplied. There seems to be an exception in Ce salam hanthey
mul.e pwa (’na) poca ‘Let’s try asking that fellow’ and similar expressions, so you will find a “ + ” at
the intersection “M-M” in the chart. A more involved sentence such as ?Cesacen ul pwa pwa poca, if
possible, is better taken as a conjoined sentence with two main verbs po- and translated ‘Let’s try
looking in that dictionary to see’.
Chart of double infinitive-auxiliary conversions
-e +
A
B
C
D
E
F
G
H
I
J
K
L
M
N
0
P
Q
R
S
T
U
V
w
X
tay-
A
-
+
+
+
+
+
+
+
+
+
+
-
+
-
+
+
+
+
+
+
+
-
-
+
chi-
B
-
+
-
-
-
-
-
-
-
+
-
+
-
-
-
-
-
-
-
-
-
-
—
-
ssah-
C
+
+
-
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
-
+
ppa ci-
D
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
+
-
-
-
peli-
E
+
+
+
+
-
+
+
+
+
+
+
+
+
-
+
+
+
+
+
+
+
-
-
+
twu-
F
+
+
+
-
+
-
+
-
+
+
+
+
+
+
+
+
+
+
+
+
+
-
-
+
noh-
G
+
+
+
-
+
+
-
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
-
+
na-
H
+
-
-
+
-
-
-
-
+
+
+
+
+
-
+
+
+
+
-
+
-
+
-
+
mek-
I
+
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
nay-
J
+
+
+
-
+
+
+
+
+
-
+
+
+
+
+
+
+
+
-
+
+
-
-
+
ka-
K
+
+
-
+
+
+
+
+
+
+
-
+
+
+
+
+
+
+
-
+
+
-
-
-
0-
L
+
-
-
+
-
+
+
+
+
+
-
-
+
+
+
+
+
+
+
+
+
-
-
-
po-
M
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
-
-
+
poy- (poi-)
N
-
-
-
+
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
+
cwu-
0
+
+
+
-
+
+
+
-
+
+
+
+
+
+
-
+
+
+
+
+
+
-
-
+
tuli-
P
+
-
-
-
+
+
+
-
+
+
+
+
+
+
+
-
+
+
+
+
+
-
-
+
pachi-
Q
+
-
-
-
-
-
+
-
-
+
-
-
-
-
+
+
-
-
-
-
-
-
-
-
talla / tao / tako
R
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
-
+
+
+
-
-
+
ha-
S
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
+
-
ttuli-
T
-
+
-
-
-
-
-
-
-
+
-
-
-
-
-
-
-
-
-
-
-
-
-
-
ci-
U
+
+
+
+
+
+
+
+
+
+
+
-
+
+
+
+
+
+
+
+
-
+
-
+
cwuk.keyss.ta
V
+
-
-
-
-
-
-
-
-
-
+
+
+
+
+
+
+
+
+
+
+
-
-
+
kaciko
w
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
-
-
+
iss-
X
-
+
-
+
+
+
+
+
-
+
-
-
-
-
+
+
+
-
-
+
+
-
-
-
A
B
C
D
E
F
G
H
I
J
K
L
M
N
0
P
Q
R
S
T
u
V
w
X
8.0. Stems.
We describe the verb stems of Korean in terms of conjugations, sets of stems that differ from each
other in shape when attached to various groups of endings. Behind the complications of the modern
conjugations you will see a simpler system at work in earlier forms of the language.
8.1. Conjugations.
Each inflected form consists of a stem + an ending. It is possible to classify stems and endings
into groups according to the ways in which alternant shapes are attached to each other. We find two
kinds of ending: in rough terms, ONE-SHAPE endings (-ko, -ta, -ci, -sey, -nun, -keyss-, -ess-, -kka,
-%, ...) and two-shape endings (-sup.nita /-p.nita, -so/-o, -una/-na, -umyen/-myen, -nunta/-nta,
-un/-n, —ul/ —1, -um/-m, -usi-/-si-, ... ). The two-shape endings have one shape which is attached to a
stem that ends, in its basic form, with a consonant and another shape which is attached to a stem
that ends with a vowel in its basic form, but some of the vowel stems have an 1-extension before
certain of the endings. The one-shape endings may actually have more than one shape - the infinitive
A Reference Grammar of Korean
PART I 231
has several shapes (-e, -a, zero, ... ) - but the choice of alternants is not correlated with the kind of
sound at the end of the stem.
With this in mind we can set up conjugations, or classes of verb stems. Those stems which
attach the shapes -sup.nita, -so, -una, -umyen, -un, -ul, -um, -usi-, etc., are CONSONANT stems.
Those which attach the shapes -p.nita, -o, -na, -myen, -nta, -n, -1, -m, -si- .are vowel stems.
Vowel stems and consonant stems attach one-shape endings, such as -ko, -ta, -ci, -nun, -%, ... , in the
same way.
In addition to the major dichotomy of consonant stems and vowel stems, we also find a few h-
dropping stems that we will call ambivalent, and several verbs with minor irregularities that we can
lump together as irregular stems. Among consonant verbs we distinguish:
(1) stems ending in Ihl which are regular in the Hankul orthography (and if -ah- take the infinitive
ending as -a) but ambivalent in colloquial speech except for the infinitive, which is regular -e (but -a
after -ah- or -oh-) - Cf the infinitive --ay of the truly ambivalent stems, such as kuleh- ‘be so’.
(2) stems ending in /w/ and /!/ which are quite regular in our Romanization - though they show
automatic alternations unusual outside the inflectional system (/w/ -* Ipl before a consonant, and /!/ -*
It I) - but which must be treated as the special classes of “irregular P, irregular T” in the analysis
underlying the Hankul spelling, reflecting the fact that the alternation is the result of a pre-Hankul
lenition of p and t before a stem-final *%- which was later elided.
(3) s-dropping stems, which involve a special type of alternation that is the result of a pre-Hankul
lenition of s before a stem-final *%- which was later elided. Among the vowel verbs we distinguish
the subclasses of 1-extending vowel stems, 1-doubling vowel stems, and 1-inserting vowel stems.
8.2. Consonant stems.
Below are examples of all occurring types of consonant-final stems.
ip- ‘wear’
noph- ‘be high’
eps- ‘be nonexistent’
tat- ‘close’
kath- ‘be alike’
wus- ‘laugh’
iss- ‘exist; stay; have’
chac- ‘look for, find’
coch- ‘follow’
mek- ‘eat’
takk- ‘polish’
ilk- ‘read’
halth- ‘lick, taste’
palp- ‘tread on’
ulph- ‘intone, chant’
talm- ‘resemble’
kam- ‘shampoo’
an- ‘embrace’
anc- ‘sit down’
noh- ‘put’
ilh- ‘lose’
kkunh- ‘cut’
kakkaw- ‘be near’
tut- / tul- ‘hear’
ci(s)- ‘build’
One anomalous adjective stem ends in -lw-: selw-, a contracted form of selew- ‘sad’. And a
truncation of ilu- ‘be early’ creates the anomalous stem ilq- that is found in what seems to be a dialect
form, ilqkena malkena = ilukena nuc.kena ‘early or late, sooner or later’. A comprehensive list
of the shapes of shorter stems is arranged by conjugation in Appendix 1. Some of the consonant-final
stems turn up in dialect versions with a final -u-: simu- for sim- ‘plant’ (see the remarks at the end
of §8.3.1), siphu- for siph- ‘desirous’, kathu- for kath- ‘like, same’.
8.2.1. Stems ending in sonants.
A typically voiceless obstruent (p t s c k) is reinforced ( -* pp tt ss cc kk) after a stem-final m, n,
or 1 (reduced from syllable excess, since a basic 1 is treated as Itl in this environment, §8.2.4). That
reinforcement, not present in certain dialects and fairly recent in the central area (see Martin ? 1991), is
ignored by the Hankul spelling, but we note it in our Romanization by a dot. To be consistent, we
would like to use “-q” in place of the dot, but that would be misleading in terms of the Hankul
spelling. The reinforcement is completely automatic only within the inflectional system, though it is
232 PART I
A Reference Grammar of Korean
widespread in other parts of the structure (especially after /I/), where we write it with q; see §1.5.
Examples:
-ko
-ta
-ci
-sup.nita
-nun
halth-
halth.ko
halth.ta
halth.ci
halth.sup.nita
halth.nun
‘lick’
/halkko/
/haltta/
/haled/
/halssumnita/
/hallun/
nam-
nam.ko
nam.ta
nam.ci
nam. sup.nita
namnun
‘remain’
/namkko/
/namtta/
/named/
/namssumnita/
/namnun/
talm-
talm.ko
talm.ta
talm.ci
talm.sup.nita
talm.nun
‘ resemble’
/tamkko/
/tamtta/
/tamed/
/tamssumnita/
/tamnun/
an-
an.ko
an.ta
an.ci
an.sup.nita
a mum
‘hug’
/ankko/
/antta /
/ancci/
/anssumnita/
/annun/
anc-
anc.ko
anc.ta
anc.ci
anc.sup.nita
anc.nun
‘sit’
/ankko/
/antta/
/ancci/
/anssumnita/
/annun/
In 15th-century Korean one-syllable stems that ended in -m- (including -lm-) and •••«-, like the
counterparts of the modern -(s)-, -W-, and -TIL- stems, began with the rising tone marked by the double
dot (“) when attached to endings that began with a vowel - specifically, the infinitive the
modulator -'w%-, and (interestingly) the honorific -%'si-. But before other endings, the stem began
with the low tone that is left unmarked, and a high pitch (the single dot ') appears at the beginning of
the ending. From that we conclude that these stems were originally dissyllabic -m%-, - 'n %-, and
(before the lenition) - 's%-, - p %-, and - 7%-. The rising tone results from blending the basic
initial low with the high tone left stranded when the vowel was elided. Accordingly, a form like
"sa'mosi'm ye (1482 Kum-sam 2:3b) contains "sam- < *sa'm[o]- ‘make’ + -o'si- (honorific) but a
form like sa'mo.n i ’’la (1459 Wei 2:27b) contains the original stem sa'mo- + -n, the modifier ending
- here entering into an extended predicate with the postmodifier 'i and the abbreviated form of the
copula indicative attentive i'la, which is ’y'la but automatically suppresses the glide after lil. Forms
for ‘hug’ include the infinitive an'a (1449 Kok 57, 1459 Wei 8:85b) and the honorific modifier found
in "an'osi'n i ’-ngi ’’ta (1459 Wei 8:86a; sic, "an-'o-) ‘hugs’, which has a rising accent like the
gerund "an'kwo (1459 Wei 8:100-1) and the deferential infinitive "anzo'Wa (1449 Kok 23), among
other forms. An explanation similar to that for -m- and -n- stems accounts for some ofthe peculiarities
of the -1- verbs (§8.3.2), which were originally •••/- stems.
8.2.2. Stems ending in h.
When attaching an ending that starts with a consonant, a stem that ends in a vowel 4- h treats the
h as /t/ - which is then subject to automatic alternations (§2.6) - unless the ending-initial consonant
is t, c, or k, with which the h undergoes metathesis. A stem that ends in a sonant 4- h (namely lh and
nh) drops the h unless the ending-initial consonant is t, c, or k, with which the h undergoes
metathesis. But stems which end in vowel 4- ph, th, or ch reduce the syllable-excess to p or t before
attaching a consonant-initial ending of any kind, and show the aspiration only when the ending begins
with a vowel and so can accommodate the syllable excess. (There are no endings that begin with p;
there are no verb stems ~kh-.)
-ko
-ta
-ci
-sup.nita
-nun
noh-
noh.ko
noh.ta
noh.ci
noh.sup.nita
noh.nun
‘put’
/nokho/
/notha/
/nochi/
/nossumnita/
/nonnun/
ilh-
ilh.ko
ilh.ta
ilh.ci
ilh.sup.nita
ilh.nun
‘lose’
/ilkho/
/iltha/
/ilchi/
/ilssumnita/
/illun/
kkunh-
kkunh.ko
kkunh.ta
kkunh.ci
kkunh.sup.nita
kkunh.nun
‘sever’
/kkunkho/
/kkuntha/
/kkunchi/
/kkunssumnita/
/kkunnun/
A Reference Grammar of Korean
PART I 233
Before a vowel, the stem-final h goes unpronounced but it is sometimes restored for emphasis, as
described in §2.7.3, or as a reading pronunciation. In Middle Korean a noun, as well as verbs and
adjectives, could end in a basic h (nwo h on ‘the cord’), which metathesized when a particle such as
'two or 'kwa was attached, and the resulting strings were written phonemically: nwo'khwa *- nwoh
'kwa ‘with the cord’. Before a voiced consonant or as a free form, such nouns suppressed the final h
and it went unwritten in Hankul: nwo nwoh ‘cord’. For a list of these nouns, see §4.2. There are a
few compound nouns that may show the metathesis, such as pol'thwok (1459 Wei 21:7b; 1466 Kup
1:36a; 1481 Twusi 22:13a, 16:24a) ‘elbow’ ?< polh twok, treated by LCT as pol[h] 'thwok but the
second noun appears only in this compound.
According to the rules in KEd a stem that ends -lh- is treated like -1-, but that will not account
for the fortition that is found in ilh.sup.nita /ilssumnita/ and ilh.so /ilsso/. Kim-Renaud (1986:24-5,
22:n7) would derive such forms by three rules: ilh-so -* *ilt-so (“h-unreleasing”) -*■ *ilt-sso (“post¬
unrelease fortition”) -*• ilsso (“coronal [= apical] deletion”). I wonder, however, if the fortition may
not be simply a part of the almost automatic rule Is -*■ lss (§2.6), which applies generally to most
strings, with a few exceptions (in Seoul) such as -I se, in which the particle can be treated as an
abbreviation of ey se. Notice that the KEd rule implies that the reductions -lh -* -1, together with
reductions from -1th-, form a new category, not to be confused with either the -T/L- stems (ket-/kel-
‘walk’) or the -L- stems (ke-1- ‘hang’), for the modifier forms differ:
ilh- ‘lose’ ilh.sup.nita /ilssumnita/ ilh.nun /illun/
ket-/kel-‘walk’ ket.sup.nita/kessumnita/ ket.nun/kennun/
ke-1- ‘hang’ kep.nita/kemnita/ kenun < ke non
8.2.3. Stems ending in w: -w- (= -p/w-).
Stem-final w alternates with p before a consonant and coalesces with a following u in the vowel
phoneme wu; and that is one reason we write what is traditionally regarded as a single phoneme with a
digraph that consists of w + u. The basic w is heard only before the infinitive ending -°/a and the past-
tense element derived from the infinitive (-%ss-), yet also sometimes before the derived adverb/noun
ending -i: tewi ‘warmth’ but kakkai ‘vicinity, nearby’. The following examples show the contrast
between a vowel stem ending in -wu (‘give’), a -p/w- stem (‘help’), and a p-final stem (‘wear’):
-ko
-sup.nita
-nun
- e /a
-una
cwu-
cwuko
cwup.nita
cwunun
cwue 1
cwuna
tow-
topko
topsup.nita
top.nun
towa
towuna
ip-
ipko
ipsup.nita
ip.nun
ip.e
ipuna
1 Usually not distinguished in pronunciation from /cwuwe/ ‘pick up’ (§2.7.6).
The modern Hankul writing system has no way to show a w at the end of a syllable, since Iwl is
written as part of the vowel medial, and so the infinitive towa has to be written to + wa and the
adversative towuna must be written to + wu + na, even though the endings are clearly -a and -una.
Since morph-final w occurs only for inflected stems, the alternations of w are completely automatic -
phonemically determined. (See also the remarks on -W- stems in §9.5-6.) In some of the provinces the
-w- stems are treated as regular — p- stems: Hwanghay and Phyengan (Pak Wensik 25); Kyengsang,
Cenla, and Hamkyeng (Choy Hyenpay 1959:332-3); Mkk 1960:3:33 cites Cincwu in South Kyengsang.
The w is thought to have been a voiced fricative in the 15th century (/W/), and the stems were
probably lenited from an original *-p(%)- under conditions that kept them apart from the -p- stems,
where the labial stop did not lenite (see Martin 1973, Ramsey 1975).
There are many adjective stems that end in w, but only a few processive verbs: kiw- ‘mend,
darn’, kwuw- ‘cook, broil’, nwuw- ‘lie down’, nwuw- ‘bleach’ [rare except in the passive nwui- =
nwl- ‘get bleached’], poyw- ‘(I) humbly see’, tow- ‘help’, yeccwuw- ‘tell (a superior)’, cesswuw-
[obsolete?] ‘bow to a divinity’ = cesswu(s)- [dialect?], cwuw- ‘pick up’, tut-caw- [literary, archaic]
‘(I) humbly hear/listen’.
234 PART I
A Reference Grammar of Korean
The anomalous adjective selw- is contracted from selew- ‘be sad’; compare the noun selwum
‘sadness’ < selw- + -um (substantive).
Middle Korean treated these -P/w- stems in much the same way. The w was also written as W (the
voiced bilabial fricative) or as Ww. The stem "twoW- ‘help’ seems to be unique in that it is often
spelled "two[W]-, with the labial consonant/ semivowel elided but the endings attached as if it were
still there. That is similar to what happens for the -(s)- stems, once the MK -z- is totally lost, as in
modern dun < ci'zun ‘ ••• that has built’. Examples of ‘help’: two'f ]a (1463 Pep 1:14b, 1481 Twusi
8:50b); "twof Jo'sya (1462 ’Nung 1:37b); two'f ]ol (1462 'Nung 8:57a), twof J'o.l i ye (1465 Nay
3:62a), two 'f Jon (1482 Nam 1:65b). If a -w- were inserted before the epenthetic o of all but the first
example, the forms would be taken as two 'wo-, the modulated stem. Compare this stem with "kwoW-
‘pretty’: kwo'wa (1481 Twusi 22:43a), "kwoWo'sya (1459 Wei 21:211b). The stems with other vowels
have either W or w in the forms that do not have p\ "swuyW-, "kiW-, mu'zuyyeW-, sa'wonaW-, ...; also
••• IwoW- and -• loW-. When the adverb-deriving suffix -i is attached, however, any -w- stem normally
drops the final labial: "swuy'i ‘easily’ < "swuyW-, "kwo'i ‘nicely’ < "kwoW-, ... . Interestingly, King
199lb:7 reports that Soviet materials on Korean from the 1920s consistently preserve the unlenited p
for the -P/w- stems, treating them as regular -p- stems, but top- ‘help’ is given exceptional forms with
the labial consonant eroded: toa for the infinitive (= towa) and towum for the substantive.
Roth (1936:163) gives as the “extended” stem for kwup- ‘bake’, cwup- and tip- ‘pick up’ the
forms kwu-u-, cwu-u-, ci-u- without the labial glide, but (173-4) he has top- ‘help’, kop- ‘be pretty’,
nwup- ‘lie down’, and etwup- ‘be dark’ with the extension (p -») -wu, as expected, except that cip- is
nonleniting in the standard language. His data may reflect dialect divergences.
The MK stems began with the rising pitch (") except when attaching endings that began with a
basic vowel. We explain that by assuming that they were originally low-initial dissyllables which had
retained the high pitch of an elided vowel ••• p[ U/ of- (before lenition of the p to W) and blended it with
the initial low to produce the rise. For more on this, see the discussion in §8.2.1.
8.2.4. Stems ending in l: -T/L-.
When attaching an ending that starts with a consonant, a stem that ends in a basic 1 treats that
final consonant as if it were It/. Korean dictionaries cite verbs in the indicative assertive form -ta, and
that form has an orthographic -t- instead of -1- for these stems; Korean grammarians treat the stems as
“irregular T stems”, just as they treat the -w- stems as “irregular P stems”. Pak Wensik (24) says the
stems are treated as regular t-final consonant stems in Hwanghay and Phyengan. (Kim Yengpay
1984:53 says tut- ‘hear’ is the only stem of this type that is regular in Phyengan; sit- / sll- ‘load’ and
the others all lenite the t to 1 as in Seoul.) Mkk 1960:3:33 says there are two variant treatments in
Cincwu in South Kyengsang: the local mul.ko /mulkko/ and the standard mut.ko /mu[t]kko/ ‘asking’.
The Cincwu version, which may be regarded as —lq-, is also reported from Hamkyeng. The existence
of the lq may lend support to Cook’s speculation that these stems have an underlying form -It-,
though I believe that to be historically inaccurate. An alternative theory has the -T/L- verbs continuing
a lost distinction between two kinds of liquid: earlier *r as well as I. The most likely historical
explanation, however, is that these stems have lenited a final t (underarticulating the stop as a flap)
under conditions that kept them apart from the -t- stems that did not lenite (see Martin 1973, Ramsey
1975), so that the -T/L- stems are quite parallel to the -w- (“p-leniting”) stems. Dialects which do not
differentiate these two types of stems from their nonleniting counterparts, the -p- and -t- stems, just
never underwent the lenition. Those dialects which lack the -t- version altogether have generalized the
lenited forms, and that accounts for the data reported by *Yi Iksep from Myengcwu county of
Kanglung, where kelkkwu means ‘walking’ (standard ket.ko) and kelkwu means ‘hanging it’ (standard
kelko), kelumwun means ‘if one walks’ (standard kel.umyen) and kelmun means ‘if one hangs it’.
The consonant stems ending in 1 (lenited from t) are to be kept distinct from 1-extending vowel
stems (§8.3.2), for these are treated as “regular”, i.e. 1-dropping, stems by the Korean grammarians,
who fail to draw the major dichotomy between consonant-final and vowel-final stems that we rely upon
here. The grammarians are historically correct, in that etymologically the extension is part of the stem.
A Reference Grammar of Korean
PART I 235
Interestingly, the 1930 grammar of the Soviet Koreanist 0 Changhwan treated the -L- stems much as
Martin 1954 and this book do (King 1991d). The following examples illustrate the differences between
an 1-extending vowel stem (ke-1- ‘hang’), a consonant stem that ends in 1 (kel- ‘walk’), and a regular
consonant stem that ends in t (ket- ‘gather up, fold/roll up’).
-ko
-sup.nita
-nun
-una
ke-1-
kelko
kep.nita
kenun
kel.e
kena
kel-
ket.ko
ket .sup.nita
ket.nun
kel.e
kel.una
/ke[t]kko/
/kessumnita/
/kennun/
/kele/
/keluna/
ket-
ket.ko
ket.sup.nita
ket.nun
ket.e
ket.una
/ke[t]kko/
/kessumnita/
/kennun/
/kete/
/ketuna/
The only form in which the 1-extending vowel stems fall together with the 1-final consonant stems
is the infinitive and the past-tense forms that are built on it: kel.ess.ta ‘hung’ or ‘walked’.
Variations of /!/ with III turn up spottily in other parts of the vocabulary, as in these words:
ithul ‘two days’, ithut nal /ithunnal/ ‘the next day’
meychil, meychit nal /meychinnal/ ‘what day of the month’
puchwul ‘squatting board in toilet’, puchwut tol ‘squatting stones’
phul ‘grass’, phut so /phusso/ ‘cow on summer diet of grass’
sel (nal) ‘New Year(’s day)’, set tal ‘December’
sahul ‘third day’, sahut nal /sa(h)unnal/ ‘third day of the month’
swul ‘spoon’, swut kalak /swukkalak/ ‘spoon’
ca-1- ‘be fine, small’, cat-talah- ‘be quite fine (small)’
se-1- ‘be unfamiliar’, set-pulu- /seppulu-/ ‘be awkward, clumsy’
He Wung (313:n35) explains set tal, swut kalak, and mus - = Imutl (1722 Sipkwu) = mwul(-i) ‘the
(whole) group, all’ < mwul as a /t/ that comes from the Isl of se[l] s tal < se/lj s "tol (1466 Kup),
swu[l] s kalak, and mu[l] s - , used to represent the adnominal (genitive) particle (Cf Part II, s Note
1). That leaves sel < "sel (1481 Samkang) and swul < 'swul (1481 Twusi) as the original forms. In
KEd I treated swul as a further contraction of swut kal(ak) and did not try to explain sel, though it
would have been possible to suggest the contraction se[t ta]l. Those explanations, however, must be
rejected in favor of He Wung’s, despite my remarks in KM 54:n32, and his explanation will apply to
the other nouns above, as well. That is, all noun-final alternations of -1 and -t- are from -[1] s ••• ,
with the liquid elided (for more on that, see §2.11.3 and the entry for the particle s in Part II). Notice
that "sel also meant ‘years of age’ (modern sal) and therefore may be connected in some way with
"sal- ‘live, be alive’.
See Martin 1983:27 for etymological examples of lenited t -* l. In some cases the source of the
lenition survives as the modern affricate because of the (southern) merger ty > c: tolyang or tocang
‘Buddhist seminary’ < MK "twotyang < "two- "TYANG. The lenited form of the plain copula ila *-
i-ta is used in quotations.
A note on notation. Contrary to the historical development, we are not considering the -1- of the 1-
extending bases as elided (dropped) in the several paradigmatic forms where it fails to appear, and
therefore do not use an apostrophe to mark its absence. We write santa for ‘lives’ rather than
“sa’nta”. Accordingly, we will not consider as abbreviation the unextended form of the stem in verb
compounds such as e-nok- (< e[l]-) ‘freeze and then thaw’ and tu-nallinta (< tu[l]-) ‘ lifts it and
makes it fly’. If we modify our description to match the history, we could still omit the notational
reminder of the elision, which goes unmarked in the Hankul orthography, saving our apostrophes for
more meaningful cases of optional contractions, such as l’ for lul, and elision of the final consonant of
a noun in compounds (pu’ son ‘fire scoop’, mu’-tepta ‘is sultry’, na’-nal-i ‘day after day, daily’),
rather than let them clutter up the verb paradigms.
The stem tut- / tul- ‘listen, hear’ is the only -T/L- stem with the vowel lul, and in MK it was
unique in having the same low initial accent in all forms. The r-leniting stems of Middle Korean that
did not have the minimal high vowels u or o began with the rising pitch (") except when attaching
236 PART I
A Reference Grammar of Korean
endings that began with a basic vowel. (The expected rising pitch automatically lost its high component
in closed syllables like C%C, leaving only the low pitch: ["] C%C- < *C% C/%/-.) We explain that
by assuming that they were originally low-initial dissyllables which had retained the high pitch of an
elided vowel - t[ u to]- (before lenition of the t to 1) and blended it with the initial low to produce the
rise. For more on this, see the discussion in §8.2.1.
8.2.5. S-dropping stems: -(s)-.
A few stems end in basic s when a consonant-initial ending is attached but drop the s when the
ending starts with a vowel. The selection of the vowel-initial alternant of a two-shape ending is just
like that for any other stem so that shapes beginning -u- will often follow the vowel that remains
when the s drops. In ordinary speech, however, the minimal vowel u is often dropped after a vowel
(§2.7.5) and that leaves the stem shape much like that of the h-dropping ambivalent stems and the
regular h-final stems that drop the h between vowels (§8.2.2, §2.7.3). It will prove helpful to compare
a simple vowel stem (na- ‘emerge’), an h-final consonant stem (nah- ‘give birth to’), an s-dropping
stem (na(s)- ‘get/be better’), and a regular s-final stem (as- ‘seize’). The table below shows the
pronunciations heard for each spelling.
na-
nah-
na(s)-
as-
-ko
nako
nah.ko
nas.ko
as.ko
/nako/
/nakho/
/na[t]kko/
/a[t]kko/!
-ta
nata
nah.ta
nas.ta
as.ta
/nata /
/natha/
/natta/
/atta/
-ci
naci
nah.ci
nas.ci
as.ci
/naci/
/nachi/
/nacci/
/acci/
-sup.nita
nap.nita
nah.sup.nita
naup.nita
as .sup.nita
/namnita/
/nassumnita/
/naumnita/ -»
/assumnita/
nah.up.nita
/namnita/
[dialect]
/naumnita -»/
/namnita/
-nun
nanun
nah.nun
nas.nun
as.nun
/nanun/
/nannun/
/nannun/
/annun/
-%
na
nah.a
naa
as.e(, asa) 2
/na/
/naha/ -»
/naa/ = /na/
/ase(, asa)/
/naa/ = /na /
-una
nana
nah.una
nauna
as.una
/nana/
/nahuna/ -*■
/nauna/ -*■
/nauna/ -*■
/nana/
/nana/
1 The bracketed t is normally suppressed (§2.6): -tk- -» -tkko -* -kko.
2 The Hankul orthography standardizes the historical spelling as.a, and in certain
dialects only that form is heard, but we have chosen as standard the Seoul as.e (see
§9.4), though the verb is uncolloquial.
The pronunciation /nassumnita/ corresponds to both nass. s up.nita ‘emerged’ ( *- na-, past formal)
and nah.sup.nita ‘gives birth to’ ( *- nah-, nonpast formal). The plain past of na- is /natta/ (nass.ta)
with short vowels; the long first-syllable vowel in /natta/ signals ‘gave birth to’ (nah.ass.ta) or ‘got
better’ (naass.ta). Notice that the shape of the infinitive after -a(s)- and -a(h)- is -a, even though
Seoul speakers use the shape -e after -aC-, including as- and the more colloquial ppay-as-. Despite
the Hankul orthographic prescription we treat as.e as the standard colloquial infinitive and speak of
as.a as a literary or dialect version of that. Under the influence of schoolroom and dialect
pronunciations the Seoul colloquial standard may be reverting to the older form, but it would be
premature to say now (1991) that the historical change in Seoul has been reversed.
A Reference Grammar of Korean
PART I 237
With the suppression of postvocalic h and u, the h-final consonant stems fall together with the s-
dropping stems when the infinitive or a two-shape ending is attached, and together they differ from a
simple vowel stem (such as na-) only by the length of the vowel left behind. But the forms are kept
distinct in Hankul spelling, at least by those who spell correctly, since the spelling is based on the
uncontracted forms.
It will be recalled that long vowels in the 1-extending, s-dropping, and -w- stems are shortened
before endings beginning with A vowel (§2.7.2). In general, that is not true of the h-final
consonant stems: coh.a ‘is good’ = /coa/ or, with epenthetic glide (§2.7.6), /cowa/. But occasionally
the shortened forms are heard: /co(w)a/ or even /cwa/ and (§2.7.4) in rapid speech leal. In /cona/ *-
/co(h)una/ for coh.una ‘is good but’ we could attribute the length to the dropped -u- but we might just
as well say it is retained from the basic shape. The shortening of the long vowels reflects their origin
as the MK low-high rising tone which resulted from contraction of the high-pitched syllable following
the basic low with which the stems began (see below).
According to Mkk 1960:3:33, speakers of Cincwu in South Kyengsang treat only the stem ci(s)-
‘build’ as s-dropping; the other s-dropping stems are regular s-final consonant stems like as- ‘seize’
and wus- ‘laugh’. Horne 1950-1 came across only the one stem cl(s)- and decided on a clever analysis
that treated the stem as basic ciy-, perhaps misled by the y-epenthesis (§2.7.6; §2.11) that makes the
infinitive cie sound like /dye/. Finding cases with other vowels, I amended her analysis and treated
these stems as ending in a basic q, to account for the reinforcement reflexes that are the only evidence
in Seoul speech that the stem has a basic final consonant. In the 15th century the stem-final consonant
was a voiced [z] before a vowel, so that the infinitive was spelled ci'ze, using the obsolete triangle
symbol for the MK Izl. Since modern dialects retain forms such as /nasa/ for naa ‘get/be better’ *-
na(s)- and lack the distinction from regular -s- stems, we assume that the MK version was an
ephemeral lenition of an earlier -s-, a lenition which took place under conditions (a later-elided
minimal vowel) absent from the regular s-final consonant stems (see Martin 1973, Ramsey 1975). For
more on the history of the stems with •••($)- < -z- see §2.11.4.
The MK verb kuz- ‘pull’ is the only s-leniting stem with the vowel lul and it continues its initial
low pitch through a following lul, as in kuzu'm ye (1481 Twusi 23:10a), but that is a reduction of
*kuzGu-, as we can see from forms with other vowels, such as the infinitive kuz'Ge (1463 Pep
2:200b), and the earlier shape of the stem is reconstructed as *kusuk- (see §8.3.1). On the other hand,
the MK verb toz- ‘love’, with the other minimal vowel o, is from *to's[o]-, and has such forms as
to'zo'm ye (1462 'Nung 4:31a), to 'zon (1462 'Nung 9:96a), and to za (ibid.). The other stems began
with the rising pitch (") except when attaching endings that began with a basic vowel (whether u /o or
e >ti). We explain that by assuming that they were originally low-initial dissyllables which had retained
the high pitch of an elided vowel ••• s[ u /o]- (before lenition of the s to z) and blended it with the initial
low to produce the rise. For more on this, see the discussion in §8.2.1. Presumably koz- ‘cut’ is like
toz- ‘love’ and accordingly is from *ko's[oJ-, but only the infinitive ko'za is attested. Similarly only the
infinitive puz Ge is attested for puzG- < *pusuk- = pozG- < *posok- ‘crush’.
8.3. Vowel stems.
There are groups of stems which end in every vowel but one: swl- ‘rest’, sey- ‘count’, toy-
‘become’, nay- ‘put out’, ssu- ‘use; write’, sa- ‘buy’, cwu- ‘give’, po- ‘see’. The exception is the
vowel e, for the only examples of -e- in Seoul speech are the abbreviations ile-, kule-, cele-, and
ette-, and these are irregular stems (§8.3.5) because the infinitives (ilay, kulay, celay, ettay) are like
hay, the irregular infinitive of ha- ‘do/say /be’. The Hankul spelling writes the stem su- ‘stand’ as se-
(thereby clarifying to some extent the causative stem seywu- ‘make stand’), and that is what it must
once have been (for MK had °sye-), but in Seoul speech the verb is regularly su-. There are, however,
a few cliches which retain sen for the modifier (used as an adnoun) instead of the more usual sun. Two
stems end in -ye- in the Hankul spelling (and in non-Seoul speech), as they did in Middle Korean:
phye- ‘smooth out, ... ’ and khye- ‘turn on (lights), ... ’. I treat these as back formations from the Seoul
stems phi- and khi-, based on the contracted infinitives phye *- phie and khye *- khie, despite the
238 PART I
A Reference Grammar of Korean
earlier versions and the variant pronunciation phey- and khey-, which is apparently confined to
northern speakers. A similar case is kennu-, spelled kenne- by the Korean grammarians, thereby
clarifying the causative kenney-. The Hankul spellings are historically correct: phi- was °phye- in the
15th century, and khi- was first attested in the early 16th century as khye-\ kennu- was MK "ken'ne-
< "ket'na- (the second vowel assimilated to the first) from a compound verb "ket-°na- (‘walk’ +
‘emerge’).
Most of the endings attach to the vowel stems in a simple and expected fashion, but there are
various complexities involving both the ending and the stem for the infinitive form; they are set forth
in §9.4.
8.3.1. ^doubling vowel stems: -ll-.
The 1-doubling vowel stem has a shape which ends in vowel + lu-. When the infinitive (-e/-a) or
the past tense (-ess-/-ass-) is attached, the vowel u drops, as expected, and the remaining 1 geminates
- as not expected: pulu- -* pulle ‘calls’, molu- -* molla ‘does not know’ (the long 6 in the infinitive
and forms derived from it is an irregularity). Many Koreans regularize these verbs by doubling the 1
everywhere; they treat the stems as pullu-, mollu-, etc. Since the modern Hankul system makes no
provision for two l’s at the end of a syllable block, the second 1 is perforce written as the initial of the
second syllable (pul + le, mol + la) even though the infinitive ending itself is just the final vowel.
The odd behavior of these stems goes back to the earliest Hankul texts, so we must reconstruct a
still earlier history to account for them. The basic forms were probably pre-MK *pulul- and
*"mwolol-. When a consonant was attached, the final liquid dropped, pulu-'ta < *pulul-'ta and
pulu-'kwo < *pulul-'kwo ; but when a vowel was attached, the minimal vowel MK % ( > u) itself
dropped: pull-e < *pulul-e. Yet among the modern -11- verbs there are some, such as talu- ‘be
different’, for which the 15th-century infinitive was given a spelling (with “•••/.-”) that we interpret as
I-IG-I with the liquid followed by a consonant (probably a voiced velar fricative): talle = talla <
tal'Ga. These stems we reconstruct as pre-Hankul *•••/%£/-, probably lenited from *~l%k-, so that the
MK stem talo-ltalG- was earlier *taloG- < *talok- and the *G dropped before a consonant ( *taloG-ta
-*■ MK talo'ta > modern taluta) but before a vowel the cluster -IG~ assimilated the fricative to the
liquid and produced the modern -11-: *taloG-a > MK tal'Ga > talla > (Seoul) talle. The Taycen
version of taluta is taltha, with a stem talh- that may reflect the G (see also the Phyengan version, p.
240); another Cenla version talpu- (Choy Hak.kun 1978:1191) either carries a suffix or implies that
the reconstruction *talok- should be corrected to *talop-. (LHS gives a Kyengsang version as talp-.)
There is other evidence pointing to an original shape like *-/%p- for the doublets that lie behind
ccalp-/ccelp- (< cyelp-) < tyalo- / tyel %- < tyelG- < *tyelW- ‘be short, fine’ ?< *tyalop-ltyelup-,
nelp- and nelu- < ne/%- ‘be wide’ ?< *nelup-.
Certain peculiarities of dialect versions of stems are also to be accounted for in terms of stem
alternants in the Hankul texts for which an earlier single form is to be reconstructed. Most of the types
are represented both in verb stems and in nouns, as shown below. (This table is adapted from Martin
1982/3:8-9, with corrections. The Middle Korean alternants are followed by the modern Seoul forms.)
(1) *-mok- *simok- ‘plant’
simo-lsimk- > slm(q)-
*tomok- ‘soak’
tomk-ltomo- > tamku-
*-mwok *namwok ‘tree’
namwo / namwok- > namu
*-mwuk *kwumwuk ‘hole’
kwumwu I kwumk" > (kwumeng)
*pwulmwuk ‘bellows’
pwulmwu Ipwulmk- > (phulmu)
A Reference Grammar of Korean
PART I 239
(2) *-n %k
* s u 4)k(-) *posok- ‘ crush ’
pozo- / pozG- > pa(s)-
= paswu-
*kusuk- ‘pull’
kuzu-lkuzG- > kku(s)-
(3) *-mbk(-) *talok- 2 ‘differ’
talo-ltalG- > talu- / tall-
(4) *-/%/(-) *molol- ‘dry up’
molo-lmoll- > malu-/mall-
*hulul- ‘flow’
hulu-lhull- > hulu-/hull-
(5) *-lok,-lol
*nyen%k ‘other’ ?< *nyonok 1
nyen % / nyenk- > yenu
*asok ‘younger sibling’
azolazG- > awu
*yes u 'bk ‘fox’ ?< *yosok' L
yez u fo / yezG- > yewu
musuk ‘radish’
muzu/muzG- > muwu, mu
*calok ‘gunnysack’
calolcalG- > calwuj
*colok ‘handle’
cololcolG- > calwu 2
*nolok ‘ferry’
nololnolG- > nalwu
*nwolok ‘roe deer’
nwolo / nwolG- > nolwu
*siluk ‘steamer’
silulsilG- > silwu
*molol ‘ridge’
mololmoll- > malwu
*holol ‘one day’
*holo / holl- > halwu
*kolok, *kolol ‘powder’
kololkolG- , Ikoll- > kalwu
1 When the vowel of the first syllable is e, the reduced vowel of the second syllable appears
both as u and as o, so we reconstruct an undecided *%. The situation probably points to
original *yo- for such morphemes, even in the absence of other evidence.
2 Or perhaps *talop- (see above). What evidence we have for the nouns points to a velar, not a
labial: in many dialects (Choy Hak.kun 1978) ‘roe deer’ is nolki or nolkay(ngi), and
‘steamer’ is silki in Kangwen (Tokyey) and also (Kim Yengpay 1984) in Phyengan.
For the last example the evidence may indicate competing versions, one of Type 2 and the other of
Type 4, the two types represented by the modern 1-doubling stems. Dialect forms (kalgi, kalgu)
confirm the velar, while forms such as kallu and kalli may be the result of IG > //, and that is the
source of the doublet, as confirmed by the dating of kol.l oy (1795 'No-cwung 1:20b [K]) < kol’G oy
( ? 1517- 'No 1:23a) and kol'G ul ( ? 1517- 'No 1:23b), where the later version of the passage has kol.l i
(1795 'No-cwung 1:20b [K]) though LCT 9b has that as “ kol.l ol”. But Ramsey 1975:40 thinks we
should reconstruct the doublet on the basis of Ceycwu kolol (Hyen Phyenghyo 1961:116a). I wonder if
all the forms may come from a unique shape *kolGol < *kol(o)kol, conceivably a reduplication.
Of the nouns with two alternants, the shape on the left of the slash is the “free” form, used alone or
before certain peripheral particles such as 'two (highlighted focus), kwa (comitative/ reciprocal), Iwo
(instrumental), s (genitive), and sometimes non (subdued focus). The shape on the right is required to
attach the primary particles 'i (nominative), ul (accusative), '%y (genitive or locative/allative), '%y
(locative/allative), and sometimes (subdued focus), as well as the copula i- .
Some of the modern dialects retain features that more clearly point to the earlier forms. In
Phyengan, for example (Kim Yengpay 1984:88, 90, 104, 168-71), the following nouns have /Ik/ for
earlier */%&: calk(i) = calwu ‘bag’ or ‘handle’, kalk(i) = kalwu ‘flour’, malk(i) also mall(i) =
malwu ‘ridgebeam’ nangk(i) = namu ‘tree’, nolk(i) = nolwu ‘roe deer’ (attested as [norogi] =
noloki in 1898 Tayshin), silk(i) = silwu ‘steamer’. And the following nouns have /lk/ for MK IG :
elkey-pit = elley-pis ‘coarse wooden comb’, kelk(w)um = kelum ‘fertilizer’, kwulk%y = kwulley
240 PART I
A Reference Grammar of Korean
‘bridle’, melkwu = melwu ‘mulberry’, molkay = molay ‘sand’, nilkwey = iley ‘seven days’,
pelkeci = peleci ‘worm’, swulkwu = swuley ‘wagon’, s'/ u lkeng = sileng ‘shelf’, tolkaci = tolaci
‘bellflower’. An earlier velar is also indicated by Phyengan alkwuy- = alii- ‘let know, inform’ and
probably taluh-/talu- ‘be different’ (with such forms as taluh.key and talun). Perhaps also (Kim
Yengpay 1984:103) the Phyengan stem kilk- = kll- ‘be long’; notice kil'li (1586 Sohak 4:53b) ?<
*kil'Gi ‘for a long time’ (= ki ll > kil.i), also MK kilh ‘path’ (? < ‘length’) and the peculiar
spelling kilq'h ol for one occurrence of its accusative (1462 'Nung 6:80a).
The noun mi 'lu ‘dragon’ looks as if it should have the forms milG- < *miluk or mill- < *milul
but the only examples are of the free form in glossaries. In other texts the Chinese word lywong is
used for dragon. The following words, probably nativized Chinese nouns containing the diminutive
suffix -zo (= "CO), have only one shape, unlike the words in (2) above:
co'zo > couy > cawi ‘kernel’ or (' nwun s ~) ‘pupil (of eye)’
pwo'zo > pwo[z]o > po ‘small bowl’
swu'zo, swu'zu > swufzjo ‘a seal-ribbon’ ?< *'sywuy-"CO
sway'zo > safzjo > (cwu)-sawi ‘dice’
8.3.2. L-extending vowel stems: -L-.
The l-extending vowel stem selects the appropriate alternant of a two-shape ending in the same
way as an ordinary vowel stem, but it adds an 1 to the stem before certain endings:
before all ONE-shape endings that begin with a vowel or with a voiceless consonant other than s -
and 1 Yi Ungpayk 1961:499 prefers the extension before s (of a one-shape ending), too, treating nolsey
as standard and nosey as variant;
before only those Two-shape endings that begin with I or m + vowel or y, such as -lye, -la, -myen
- and -m when it is followed by a vowel-initial particle shape such as ulo, ey, or a.
Perhaps the rule can be more clearly stated: the -1- extension is present except before p, s, n, mC,
ml and except before the ending -o. Accordingly the stem tu-1- ‘enter’ makes the forms tulko, tulta,
tulci, tulkeyss.ta, tul.e, tul.ess.ta, tullye, tulla, tulmyen, tulm ulo; but tuo, tunun, tunta, tuna, tun,
tul(q), and /turn/ (spelled tulm) + pause or consonant. A long vowel in a one-syllable l-extending
stem is shortened in the infinitive and past forms: ke-I- ‘hang’ becomes kel.e, kel.ess.ta. (The length
in the stem reflects the monosyllabification of a pre-MK dissyllable: *ke'lu- > "kel-lke'lu-.) There is
a substandard variety of speech that inserts /u/ between the liquid and a following m or 1 (also n? -
see below): pulumyen = pulmyen ‘if it blows’ (Cf An Sangchel 1988:153), wulul(q) = wul(q) ‘to
cry’ (Kim-Renaud 1986:112, who says these forms are the more common type in Chwungcheng).
According to ’Yi Tongcay 1989:147 forms like sal.umyen for salmyen, salu.Iye for sallye, and
sal.ul(q) for sal(q) (to live’), which are “less commonly used and less readily accepted” occur (for
Seoul) “only in the speech of some, mostly young, speakers”. There may also be speakers who drop
the liquid and say wumyen for wulmyen ‘if one cries’ (Kim-Renaud 1986:113:n8). In Taycen it is
usual to say mel.un for men ‘(that is) distant’ and kel.un for ken ‘(that has) hung’. In describing the
variants in conjugation of stems such as ki-l- (‘be long’) one suggestion (1936 Roth 141) is to think of
a “shortened stem” ki- and an “extended stem” kilu- as supplementary to the normal stem. That may
be helpful in considering dialect variants such as aci = alci and alusio = asio (‘know’).
It is unclear whether all of the several nonstandard treatments are analogical innovations or
whether some are simply preserving uncontracted forms from the earlier language. The Middle Korean
treatment of the modifiers (the elided /%/ is explained below):
- l[ u /o]-n -* -n, as in "an cyen'co Iwo (1462 ’Nung 9:13a) ‘since he knew’ = an kkatalk ulo; "an
ti's ’i (1449 Kok 43) ‘as if he knew’ = an tus ’i, "a n i (1482 Kum-sam 2:2b) ‘knew’ = an kes ita;
"men tuy s hhwo-'kywow lol (1459 Wei 2:69a) ‘alien teachings from distant places’ = men tey
-l[ u /o]-no-n -* -non, as in "a 'no.n i ’ la (1482 Kum-sam 2:3a) ‘knows’ = anun kes ita
-l[ u fo]-l(q) -* -ll before i or y, as in "al l ye ("mwot ho 'l ye) (1463 Pep 3:86a) ‘can you tell me (or
not)?’ and 'NYELQ-PPAN ay tull i Gwo (1482 Kum-sam 2:13b) ‘how will they possibly enter nirvana?’
= tulq kes in ko
A Reference Grammar of Korean
PART I 241
-l[ u fo]-l(q) -*■ -l(q) (except before i or y), as in tul tt ol (1463 Pep 1:55b) ‘that you can enter’ =
tulq kes ul
There is a puzzling form annon for "a non in 1887 Scott 122; perhaps it is a mistake, or a secondary
doubling of the n for emphasis.
Before the honorific, modern Seoul uses the “shortened stem” but the earlier language had the
“extended stem”:
"a'losi'm ye (1465 Wen 1:2:3:6a) = asimye ‘knows and’
"wu'lusi'n i (1449 Kok 57) = wusini ‘cried (and)’
Three situations placed a sibilant directly after a stem, and in these forms the “shortened stem” was
used, with the / elided: tuson 'ta (1462 'Nung 5:31b) ‘does one enter?’ (emotive -so-), "azop'kwo
(1449 Kok 109) ‘humbly knowing’ (deferential -"zoW-), 'tu-'sa (1447 Sek 13:58a) ‘only entering’ (=
tule 'za). The basic form of the -L- verbs must have been •••/%- but the surface forms are often
reduced to •••/- (with loss of the vowel) or — (with total loss of the syllable). For those stems beginning
with a low pitch (the unmarked tone), the reducing syllable had a high pitch (~ 7o-) and that was kept
and blended with the initial low so that, for example, *a'lo- became "al- and "a-, but the basic accent
survived in the infinitive a'l-a, the modulated forms a'l-wo-, and the forms with the elided-initial
version of the effective forms a 'l-a- (but not in "al-'Ga- with the lenited-initial version). The critical
factor for the low initial accent is that the surface form of the attached element begins with a basic
vowel. That means we must treat the honorific as basically vowel-initial: "a'l-osi-, "wu'l-usi-. One
further point: the initial low of a stem with the vowel u (but not wu) or the vowel o (but not wo) stayed
low in all forms, so that tul- ‘lift’ has such forms as tun (1481 Twusi 8:35a), tunon (1463 Pep 4:19a),
tul(q) (1527 Cahoy 3:10b=23b), tu'ti (1463 Pep 2:173b), and tu'zopke'na (1447 Sek 13:53b), as well
as tu'le (1449 Kok 73), tu'lwum (1482 Kum-sam 3:22a), and tul'Gwo (1459 Wei 7:8a). Likewise mul-
‘bite’, nul- ‘be better’, and pul- ‘envy’ (but not "pwul- ‘blow’). There is a counterpart in the initial
high pitch of the stems tul- ‘enter’ and sul- ‘vanish’, which is retained in all forms. But the two
phenomena may have come about in different ways. Middle Korean had no word-initial syllables *"Cul
or *"Col except for stems with final consonant clusters as found in "kolp'kwo (1481 Twusi 20:22a)
‘lining them up’, "solp'kwo (1459 Wei 1:15b) ‘telling a superior’, and "tulp'kwo (1463 Pep 6:154b)
‘piercing’. The few attestations to the contrary are scribal errors: "pol'sye ( ? 1512- Pak 1:37b) must be
a mistake for pol ' sye (id. 1:5b), a variant (1462 'Nung 1:37a, 1518 Sohak-cho 8:7a) of pol 'sye (1465
Wen se:68a, 1481 Twusi 7:8a, 1483 Kum-sam 2:2b) = pol'ssye (1447 Sek 6:35b; 1459 Wei 9:36a;
1462 'Nung 3:25ab, 9:117a, 1463 Pep 4:63b) ‘already’. Therefore the accent of forms such as kol'Gwo
(1466 Kup 1:10a) ‘changes and’ or tul'Gwo (1459 Wei 7:8a) ‘lifts and’ are functionally equivalent to
that of "mel'Gwo (1459 Wei 10:23b) ‘is far and’. Even without the accentual clue we will still account
for the choice of regular, extended, or shortened stem by assuming that the original stem had two
syllables, the second of which bore a basic high pitch. Stems with the higher of the minimal vowels:
* 'tu 'lu- > tul- ‘enter’, *'su'lu- > sul- ‘vanish’;
*tu'lu- > f 'Jtul- ‘lift’, *mu'lu- > f "]mul- ‘bite’, *nu'lu- > [']nul- ‘be better’, *pu'lu- > [ ]pul- ‘envy’.
This notation presumes a stage when the low-rise was actually pronounced, but such a stage may not
have existed: the reduction of the expected low-rise to just low may have been simultaneous with the
truncation. In any event, in Part II we leave the initial syllable of these stems unmarked (i.e. low),
following the Hankul spellings. Stems with the lower of the minimal vowels:
*'ko'lo- > kol- ‘grind’, *'pho'lo- > phol- ‘sell’, *'so'lo- > sol- ‘burn it’, *'sko'lo- > skol-
‘spread it out’, *'spo'lo- > 'spot- ‘be sharp-pointed; launder; sip’, *'to'lo- > tol- ‘hang’;
*ko'lo- > [ "]kol- ‘change’, *mo'lo- > ['}mol- ‘roll it up’, *no'lo- > *["]nol- ‘fly’, *to'lo- >
f 'Jtol- ‘be sweet; weigh it’.
In citing the 1-extending stems of modern Korean, it is handy to mark them off from those
consonant stems that end in 1 (lenited from a pre-MK *t) by inserting a hyphen before the extension,
for when we write tu-1- ‘enter’ that way not only are we reminded that it belongs with the vowel-stem
242 PART I
A Reference Grammar of Korean
conjugation but that it is different from tut-/tul- ‘hear’ (§8.2.4), though the two stems have in
common the shape of the infinitive tul.e ‘enters’ or ‘hears’, and the past-tense forms based on that.
In Hankul the -1- extension should always be written with the final syllable of the stem, so that
‘enters’ and ‘hears’ are both spelled tul.e, but many Koreans violate this rule by beginning the first
syllable of the ending with the letter 1 when possible, especially in the infinitive, and such spellings
were common in all but a few of the early texts. The best way to state the Hankul spelling rule is
perhaps as follows. For the 1-extending vowel stems such as tu-1- ‘enter’, write the 1 - at the end of
the last syllable of the stem - only when it is heard, but always write the substantive as -lm. For the
1-final consonant stems such as tul- ‘hear’ (the -T/L- stems), write a stem-final 1 whenever the 1 is
actually heard, but in all other forms write a stem-final t.
Some of the 1-extending stems are confused by many Koreans with regular or 1-doubling stems in
their paradigmatic forms: we find both ecilun and the standard ecin for the modifier of eci-1- ‘be kind,
good’ and situlun alongside situn for the modifier of situ-1- ‘wither, wilt’ (as in ~ chayso ‘wilted
vegetables’ and ••• son ‘withered hand’). We find also al.um for aim ‘knowledge’, and al.un for an,
the modifier of a-1- ‘know’. I have also heard /alumnita/ = al.up.nita for ap.nita. Common variants,
usually considered nonstandard in the modern language, drop the 1 before t, 1, and especially c: aci (=
alci) mos hanta ‘can’t know’, ata (= alta) siph.i ‘as we know’. For the negative auxiliary in haca
maca (= malca) ‘(no sooner than =) as soon as one does’, the shorter form seems to be the norm. All
these variants were the usual forms in earlier Hankul texts (§2.11.2), so that the modern standard
usage shows either a restoration or a preservation of spoken versions of the language that never did
suppress the 1.
8.3.3. L-inserting vowel stems.
The l-inserting stem ends in a vowel + lu, like the 1-doubling stem, but instead of dropping the u
and doubling the 1 it adds an 1 before the infinitive or the past-tense element: nwulu- ‘be yellow’ -*
nwulule. (The orthography, of course, spells the 1 with the ending: nwu-lu-le.) There are only a few
such stems: nwulu-/nolu- ‘be yellow’, phulu- ‘be blue’, (kam)-phalu- ‘be blue’, and ilu- ‘reach’.
Compare ilule ‘reaches’ with ille (< nil'Ge ) ‘says’ or (< il'Ge) ‘is early’ from 1-doubling stems ilu-
< ni/%- / nilG- < *nil U £>G- and il u 4>lilG- < *U u 4jG-. And compare nwulule ‘is yellow’ with
nwulle ‘presses down’ < MK nwu lu- / nwull- < *nwu lul-. The l-inserting stems must be < *-/%/-,
but that is clearly attested only for ilu- < ni 'lul- ‘reach’. Taycen regularizes the verb ilu- to ilulu-, so
that ilulun kos is said for ilun kos ‘place reached’.
Aside from the few l-inserting stems, all stems that end in -lu- seem to be the 1-doubling type,
with the following exceptions, which are simply among the regular vowel stems that happen to end in
-u-: ttalu- ( -* ttale or ttala) ‘conform, obey; pour’, chilu- ( -* chile) ‘pay, disburse’, tatalu- ( -*
tatale or tatala) ‘arrive’ from a consonant stem tatal- (tatat.ta), mak-talu- ( -* mak-tale or mak-
tala) ‘(an alley) be closed at one end’, and wulelu- ( -* wulele) ‘lift one’s head, look up, respect’.
South Korean dictionaries also list salu- ‘winnow’ as regular, with the infinitive sale (or sala), but
North Korean dictionaries list it as 1-doubling, like the verb salu- ‘set afire’ with the infinitive salle
(or salla). I have been unable to confirm either version, for the common way all my informants say
‘winnow’ is khi cil ha-.
8.3.4. Ambivalent stems: -(H)-.
Ambivalent stems are treated as consonant stems that end in h (§8.2.2) before -sup.nita and
before one-shape endings not beginning with a vowel (i.e. those other than the infinitive and the past-
tense element, which are like the forms of irregular stems, §8.3.5), but as vowel stems, with the h
dropped, before two-shape endings other than -sup.nita/-p.nita. These stems are all derived from an
infinitive (-el- a etc.) + a reduced form of the irregular adjective ha- ‘be’, so their infinitives and past-
tense forms are similar to hay and hayss- (§8.3.5). Stems derived from the processive verb ha- ‘do,
say’ are not ambivalent (despite mistaken spellings by some writers), they are just irregular vowel
stems like ha-. Compare the following examples:
A Reference Grammar of Korean
PART I 243
ADJECTIVE
VERB INTRANSITIVE
stem
gerund
suspective
formal statement
adversative
substantive
infinitive
kule(h)- ‘be like that’
kuleh.ko /kulekho/
kuleh.ci /kulechi/
kuleh.sup.nita
/kulessumnita/
kulena
kulem
kulay
kule- ‘do/say like that’
kuleko
kuleci
kulep.nita
/kulemnita/
In the spoken language all -h- stems (§8.2.2) are usually treated as if ambivalent, but they have
regular infinitives: neh.e (often pronounced /ne/), noh.a (often pronounced /noa/ or even /nwa/,
§2.7.7). And they always have the alternant -a (rather than -e) after -ah-, as in tah.a (often /ta/) from
tah- ‘touch; reach’. Roth (1936:158) gives the infinitive of cokomah- ‘small’ as cokoma and of
twungkuleh- ‘round’ as twungkule but has -ay for the infinitive of the other ambivalent stems.
8.3.5. Irregular stems: ha- and derivatives.
The infinitives are irregular for the stem ha- < °ho- both as the processive verb ‘do/say’ and as
the adjective ‘be’ (but not the obsolete ha- < °ha- ‘big, much, many’), and for certain processive
stems derived from it. The infinitives of ha-, ile-, kule-, cele-, ette-, and amule- are hay, ilay, kulay,
celay, ettay, and amulay. The infinitive of ha- has the literary variant ha.ye (also ha.ya). The past-
tense forms follow the pattern of the infinitive: hayss- or ha.yess-, ilayss-, kulayss- .There are no
literary variants for the derived verbs because they are abbreviations and in formal writing they are
expanded to their models: ile hay or ile ha.ye (= ilay), kule hayss- or kule ha.yess- (= kulayss-).
Ette ‘what way’ has a variant ecce, and from that comes a derived adverb (§9.6) ecci with about the
same meaning. Notice that /eccay/ ‘how’ is an abbreviation from ecci hay (-* ecc’ ’ay). There is also
a derived adverb from the adjective ha- ‘be’, with the shape hi < hi (§9.6). And we might want to
consider sikhi- ‘cause to do’ as an irregular alternant s- of the stem ha- + the bound postverb -ikhi-
(found also in tol.ikhi-); see §7.4.
8.3.6. Irregular stems: k-inserting and n-inserting.
A few stems have two infinitive forms: the normal one is formed as expected, the special one is
used only before the command particle la. Do not confuse this word with the subjunctive attentive
ending -ula/-la, which is attached directly to the stem and produces a plain command used only in
quotative constructions or literary forms. The two are indeed etymologically related, though not quite
not as directly as the shapes seem to suggest. Notice the exclamatory use of adjective infinitive + la
and see the historical remarks in the entries of Part II. For vowel stems that end in ey, oy, ay, or a the
two command structures will sound identical: sey(e) la, seyla ‘count!’; oy(e) la (spelled “wayla” in the
Hankul orthography, Cf §9.4), oyla ‘memorize!’; nay la, nayla ‘pay!’; sa la, sala ‘buy!’. But for the
other stems a difference is heard: mek.e la, mek.ula ‘eat!’; nol.a la, nolla ‘play!’; kie la, kila
‘crawl!’; ttwie la, ttwila ‘jump!’; cwue la, cwula ‘give!’; sse la, ssula ‘write!’. In Seoul the forms for
‘stand!’ are different (se la, sula) but they are identical in the literary /dialect forms (se la, sela).
The vowel stems ca- ‘sleep’, na- ‘emerge, ... ’, ka- ‘go’, and toy- ‘become’, together with the
consonant stem iss- ‘stay’, insert a k before attaching this secondary infinitive ending: cake la, cala
‘sleep!’; nake la, nala ‘emerge!’; kake la, kala ‘go!’; iss.ke la, iss.ula ‘stay!’. The vowel stem o-
‘come’ uniquely inserts n: one la, ola. In Hankul spelling the inserted phoneme is written as the onset
of the infinitive syllable, and the string with the particle is considered by the grammarians to be an
unanalyzed ending (ka-ke-la, o-ne-la, iss-ke-la). There may be other stems that belong to this class.
According to Choy Hyenpay 1959:334-5, competing variants of the type iss.e la and iss.ke la occur
for that verb and also tul- ‘hear’ (tul.e la and tut.ke la), cwuk- ‘die’ (cwuk.e la and cwuk.ke la), anc-
‘sit’ (anc.e la and anc.ke la), and su- ‘stand’ (se la and suke la). Choy treats the -ke version as
244 PART I
A Reference Grammar of Korean
“dialect” for all these stems, but that may not be entirely accurate. I have heard ka la for kake la and
for one la we can expect to hear wa la (which I am told is a “Seoul-ipsism”). In the expression toylq
tay lo toy(ke) la ‘let what may happen happen!’ either treatment is accepted. Some people insert -k-
for all one-syllable stems ending in -a-. The -k- and the -n- come from morphemes marking the
“effective” aspect of Middle Korean.
9.0. Endings.
We describe the verb endings in terms of sequence positions (§9.1): where each ending fits when
put into a long string of endings. More detailed information is given for the specific categories in
separate sections (§§9.2-8). The modern verb system developed from an earlier scheme, which is
described in a similar fashion in §9.9. The description focuses on form and shape, but both here and in
later parts of the book attention is paid also to function and meaning.
9.1. Sequence positions.
The total number of paradigmatic endings for modern Korean is well over 400. And that number
does not include structures that are here treated as inflected form + particle (such as -e se, -e to, -e ya,
-e la; -ki ey, -ki lo; -ko nun; -ci man; -um ulo, ...) or cases of modifier form + postmodifier (such as
-tun ci, -nun ya, -ulq ka, -nun tey, ... ), or abbreviated quotative constructions (such as -ta ’nta *- -ta
hanta, -ulye ’nta *- -ulye hanta, ila ’nun *- ila hanun, ... ). Korean grammarians often lump these
construc 
Live Music Archive
Librivox Free Audio
Metropolitan Museum
Cleveland Museum of Art
Internet Arcade
Console Living Room
Open Library
American Libraries
TV News
Understanding 9/11